随着中国经济发展和城镇化的推进，地铁工程数量和规模在过去几十年里持续增长，这些工程通常涉及复杂的建设条件和高安全风险。为应对这些挑战，国家和地方制定了一系列法规和标准，形成了一套较为完善的规范体系，以提升工程规范化水平、确保工程质量并降低风险。然而，在实际应用中，现有规范的应用方式和效率仍面临诸多问题，如规范内容分散、关联性差、依赖人工处理，限制了地铁工程的高质量发展。为了优化地铁工程知识管理，研究人员开始利用信息技术如本体和知识图谱来管理和应用规范知识，但是目前仍然存在规范认知不充分、技术制约以及人工智能研究不足等问题。针对上述问题，本研究结合了深度学习、大语言模型、本体以及知识图谱等技术，对知识分析、知识提取方法、知识管理方法、知识应用四个方面展开研究。
本研究首先对规范文件及其知识建立全面认知。对于文件组成，分析了规范文件的来源，划分了规范边界，并构建了包括 143 个文件的地铁工程规范文件集。从结构、语义、词三个维度分析了规范文本的特征。基于规范构建了地铁工程知识框架，并以自上而下的方式形成多层级地铁工程知识体系。
基于上述认知，本研究开发了地铁工程规范文件的知识提取方法。针对规范文件的组成特征，设计了 TEARS 定义以及相应的成分识别与重构算法。对于条文中的专业实体，构建了 NER 数据集，基于 BiLSTM-CRF 模型对规范条文中的实体元素进行识别，并引入了 BERT 对模型进行优化，实现了高效准确的实体自动化提取。此外，还对大模型在地铁工程领域的应用方法进行研究，提出一种基于大语言模型的地铁工程规范文本 NER 方法。
为了更好地组织规范知识，本研究设计了知识管理方法。结合地铁工程规范的特征与应用需求，提出了一种新的本体构建方法。以该本体为模式层，BERTBiLSTM-CRF 模型为提取技术，从规范文件中提取知识并构建了地铁工程规范知识图谱，共包含 87469 个实体与 330318 个关系。在此基础上，提出了一种将知识图谱与大语言模型结合的知识增强模型架构。最终，本研究开发了一个地铁工程规范知识管理平台，该平台具有知识导航、知识管理与知识问答三类核心功能。
整体而言，本研究对地铁工程规范形成了相对全面的认知，提出的知识提取技术为相关研究提供了技术框架，知识管理与应用方法提高了地铁工程的知识管理水平，并具有较好的应用前景。

As China's economy has developed and urbanization has progressed, the number of subway engineering projects have continued to grow over the past few decades. These projects often involve complex construction conditions and high safety risks. To address these challenges, national and local authorities have established a series of codes, forming a relatively comprehensive code system to enhance the standardization of engineering, ensure project quality, and reduce risks. However, in practical applications, the implementation methods and efficiency of existing codes still face many issues, such as scattered content, poor correlation, and reliance on manual processing, which hinder the high-quality development of subway engineering projects. To optimize knowledge management in subway engineering, researchers have begun utilizing information technologies such as ontologies and knowledge graphs to manage and apply knowledge.However, there are still issues such as insufficient understanding of codes, technical limitations, and inadequate research in artificial intelligence. In response to these issues, this study integrates technologies such as deep learning, large language models, ontologies, and knowledge graphs to conduct research in four areas: knowledge analysis, knowledge extraction methods, knowledge management methods, and knowledge application.
This study first establishes a comprehensive understanding of the codes and their knowledge. It analyzes the sources of codes, defines their boundaries, and constructs a subway engineering code collection consisting of 143 files. The characteristics of code texts are analyzed from three dimensions: structure, semantics and words. Based on these codes, a subway engineering knowledge framework is constructed, and a multi-level subway engineering knowledge system is formed.
Based on this understanding, this study developed a method for extracting knowledge from subway engineering codes. According to the composition characteristics of the code, the TEARS definition and the corresponding component identification and reconstruction algorithm are designed. For the entities in the clauses, a NER dataset was constructed. The entities within the code clauses are identified using a BiLSTM-CRF model, and the model is optimized by introducing BERT. The optimized model achieved efficient and accurate automated extraction of entities. Additionally, the application methods of large models in the field of subway engineering were studied, proposing a subway engineering code text NER method based on large language models.
To better organize code knowledge, this study designed a knowledge management method. Combining the characteristics of subway engineering codes, a ontology construction method was proposed. Using this ontology as the schema layer, and the BERT-BiLSTM-CRF model as the extraction technique, knowledge was extracted from the codes and a subway engineering knowledge graph was constructed, containing 87,469 entities and 330,318 relationships. On this basis, a knowledge-enhancement model that combines large language models with knowledge graphs was proposed. Finally, this study developed a subway engineering code knowledge management platform, which has three core functions: knowledge navigation, knowledge management and knowledge question and answer.
Overall, this study has formed a relatively comprehensive understanding of subway engineering codes. The proposed knowledge extraction technology provides a technical framework for related research. The methods of knowledge management and application have enhanced the knowledge management in subway engineering and offer promisingprospects for future application.

公路工程领域项目实施过程中会涉及大量与工程安全有关的数据、信息或知识，其覆盖范围广且来源广泛。从覆盖范围讲，相关数据、信息或知识可覆盖公路工程的各单位工程如桥梁、隧道等，且涵盖项目全生命期；从来源讲，相关数据源或知识源可分为业务生产数据和规范文本两类，前者真实反映项目在某阶段某场景下的生产情况，后者对项目安全生产过程起约束和指导作用。然而，在传统的公路工程安全管理过程中，安全管理者对相关数据、信息或知识维持着较为粗放的处理逻辑，导致领域内对业务生产数据内容及其关联信息的系统梳理，项目实施场景的准确且直观描述，不依赖专家经验及重复解读且基于规范的安全知识梳理及管理，以及围绕项目实施场景的知识点自动输出及应用等过程具有较高需求。
面对这些需求，本课题引入了本体、知识图谱、数据模版等新兴技术，依次从业务数据梳理、实施场景本体建模、规范知识库搭建以及安全管理模板设计与实现四块内容进行本文安全管理创新研究与应用模式的方法研究及结果讨论。业务数据梳理方面，提出将相关数据划分为项目与组织、结构物、机械设备、施工环境、管理文件五个维度，并充分调研五类数据的层级内容及内部关系；实施场景本体建模方面，对五个子领域数据及一个全局本体的相关概念及内部关联进行语义建模，并融合形成公路工程领域通用的实施场景本体；规范知识库搭建方面，引入传统深度学习模型TextCNN等以及预训练模型BERT、ERNIE，对数百本规范进行实体识别、关系抽取等流程以构建领域知识图谱；安全管理模板设计与实现方面，通过路径规划、检索等过程，将领域实施场景本体自动转化为通用实施场景模板，同时引入典型的作业划分方法以自动构建数百份围绕作业元素的知识反馈模板，二者结合以供管理者在准确且完整识别项目当前作业场景模式的同时及时获取相关知识。
实例验证表明，本课题所构建的实施场景本体及其模板相对传统的施工日志等记录方式可以更加完整且准确地描述项目当前的实施场景，所构建的领域规范知识库质量较好，且基于规范知识库的知识检索、知识反馈模板构建方法可代替专家自动反馈围绕作业元素的知识。总的来说，本课题提出的安全管理模式较大程度提高了我国公路工程安全管理水平，对工程建设领域的综合管理也具有较大参考价值。

In the field of highway engineering, the implementation process of projects involves a significant amount of data, information, or knowledge related to engineering safety. These resources have a wide coverage and come from diverse sources. In terms of coverage, the relevant data, information, or knowledge can encompass various units of highway engineering, such as bridges, tunnels, etc., and cover the entire lifecycle of the project. In terms of sources, the relevant data sources or knowledge sources can be divided into two categories: production data and normative texts. The former provides an authentic reflection of the project's production status in specific stages and scenarios, while the latter serves as a constraint and guidance for the project's safety production process. However, in the traditional highway engineering safety management process, safety managers tend to employ a relatively broad approach when dealing with relevant data, information, or knowledge. As a result, there is a high demand in the field for systematic sorting of production data and associated information, accurate and intuitive description of project implementation scenarios, standard-based security knowledge sorting and management without expert experience and repeated interpretation, and automatic output and application of knowledge points around project implementation scenarios.
In response to these needs, this study introduces emerging technologies such as ontology, knowledge graph, and data templates. It conducts methodological research and results discussion on innovative safety management and application models, focusing on four aspects: data sorting, implementation scenario ontology modeling, standardized knowledge base construction, as well as design and implementation of security management template. Regarding data sorting, the study proposes dividing relevant data into five dimensions: projects and organizations, structures, mechanical and electrical equipment, construction environment, and management documents. A comprehensive investigation is conducted on the hierarchical content and internal relationships of these five types of data. In terms of implementation scenario ontology modeling, the study performs semantic modeling of relevant concepts and internal associations within the five sub-domains and develops a global ontology. This integrated ontology forms a common implementation scenario ontology in the field of highway engineering. For standardized knowledge graph construction, the study introduces traditional deep learning models such as TextCNN, as well as pretrained models such as BERT and ERNIE. These models undergo processes such as Named Entity Recognition and Relationship Extraction on hundreds of standards to build a domain knowledge graph. In terms of the design and implementation of security management template, the study automates the transformation of the domain implementation scenario ontology into a common implementation scenario template through processes like path planning and retrieval. Additionally, Tables of work classification are introduced to automatically generate hundreds of knowledge feedback templates based on work elements. The combination of these two processes allows managers to accurately and comprehensively identify the current work scenario mode of the project while timely accessing relevant knowledge.
The results of the case verification demonstrate that the implementation scenario ontology and its templates constructed in this study provide a more complete and accurate description of the current project implementation scenario compared to traditional recording methods such as construction logs. The quality of the constructed domain-specific knowledge base is satisfactory, and the methods used for knowledge retrieval and knowledge feedback template construction based on the knowledge graph can replace experts to automaticly feedback knowledge related to work elements. Overall, the safety management model proposed in this study significantly improves the level of safety management in highway projects in our country and holds substantial reference value for comprehensive management in the field of engineering construction.

随着全球能源结构转型的推进，海上风电作为可再生能源的重要组成部分，正逐渐成为能源领域的焦点。海上浮式风机因其能够在深远海域利用丰富的风能资源而备受瞩目。然而，这些风机在复杂多变的海洋环境中运行，对姿态监测和控制提出了挑战，尤其是立柱的姿态变化对于保障风机的高效运行和设备安全极为关键。
本课题基于可模拟海上浮式风机各种运行姿态的缩尺模型，深入探索了风机姿态的实时监测与分析技术。重点关注立柱姿态的精确捕捉与分析，通过构建满足弗劳德相似性的风机缩尺模型，为海上浮式风机的数字孪生开发提供了理论基础和技术支撑。利用物联网技术和数据清洗方法，本课题实现了风机姿态数据的实时采集并进行了数据清洗，确保了数据的有效性与可靠性。特别是对风机立柱的形变函数进行了深入分析与拟合，以尽可能精确地捕捉其形态变化。通过自主开发的数字孪生平台，本课题实现了风机运行状态的三维动态可视化，包括立柱的形变、风机的倾斜与旋转等关键姿态的精确模拟以及风速、风向等环境参数的实时监测与展示，显著提升了风机监测与分析的直观性和效率。该平台还支持历史数据查询与回放功能。此外，平台还具备风机姿态的手动调节功能，以及基于风环境数据的自动优化能力。
在重点关注立柱形变拟合方面，本课题采用了先进的有限元技术和数值分析方法。通过对立柱表面的多点监测数据进行分析，捕捉到了立柱在不同风浪条件下的形变特性。利用非线性回归技术对立柱的形变数据进行拟合，得到了一个描述立柱形变的精确数学模型。该模型能够反映立柱在静态载荷下的形变行为，以及其在动态风浪作用下的响应，并为其形变的精确可视化展示提供了支持与保障。
本课题的成果不仅推动了海上风电领域的发展，也为海上风电机组的智能监控与运维管理提供了新的思路和方法。通过实时监测和动态可视化技术，为海上风电的高效运行和安全维护提供了有力的技术支撑，同时也为海上风电领域的数字化转型和智能化升级奠定了坚实的基础。未来，我们将继续优化数字孪生平台的功能，探索更多与实际海洋环境相适应的监测技术，以实现对海上风电系统更全面、更精确的监测和管理，为全球能源结构的绿色转型做出更大的贡献。

As the global energy structure transitions towards sustainability, offshore wind power, a vital component of renewable energy sources, is increasingly becoming a focal point in the energy sector. Offshore floating wind turbines, in particular, are garnering attention for their capacity to harness the abundant wind resources in deep-sea areas. However, operating in the complex and variable marine environment poses significant challenges for posture monitoring and control, especially since the posture changes of the turbine column are crucial for ensuring the turbine’s efficient operation and safety.
This project delves into real-time monitoring and analysis techniques for the posture of offshore floating wind turbines, based on a scale model capable of simulating various operational postures. It focuses on the precise capture and analysis of the column posture, utilizing scale models that satisfy Froude’s similarity to lay the theoretical and technical foundation for the digital twin development of offshore floating wind turbines. By employing IoT technology and data cleaning methods, the project achieves real-time collection and cleansing of turbine posture data, ensuring the validity and reliability of the data. Particularly, the project conducts an in-depth analysis and fitting of the wind turbine column’s deformation function to capture its shape changes accurately. With a self-developed digital twin platform, the project facilitates the three-dimensional dynamic visualization of the turbine’s operational state, including precise simulations of key postures such as the column’s deformation, tilt, and rotation of the turbine, as well as real-time monitoring and display of environmental parameters like wind speed and direction, significantly enhancing the intuitiveness and efficiency of turbine monitoring and analysis. The platform also features manual adjustment capabilities for turbine posture and automatic optimization based on wind environment data.
Focusing on column deformation fitting, the project employs advanced finite element technology and numerical analysis methods. By analyzing multipoint monitoring data on the column surface, it captures the column’s deformation characteristics under various wind and wave conditions. Nonlinear regression techniques are used to fit the column’s deformation data, resulting in an accurate mathematical model that reflects the column’s deformation behavior under static loads and its response to dynamic wind and wave actions, providing support and assurance for precise visualization.
The outcomes of this project not only propel the development of the offshore wind power field but also offer new approaches and methods for intelligent monitoring and maintenance management of offshore wind turbine units. Real-time monitoring and dynamic visualization technologies provide robust technical support for the efficient operation and safe maintenance of offshore wind power, laying a solid foundation for the digital and intelligent upgrade of the offshore wind power field. Moving forward, we will continue to optimize the functions of the digital twin platform and explore monitoring technologies adapted to actual marine environments, aiming for a more comprehensive and accurate monitoring and management of offshore wind power systems, thereby making a more significant contribution to the green transformation of the global energy structure.more to the green transformation of the global energy structure.

大海洋环境对人类生产生活有着重要意义，是实现绿色、协调、可持续发展的重要组成部分。海洋中污染物的扩散模拟分析对近岸海域治理，以及海洋污染防治、预警也有着重要作用。尽管目前已经有很多针对相关领域的研究和分析系统，但大多都只适用于小范围时空下某种单一污染物的扩散模拟，并且各系统间存在信息壁垒现象，难以实现海洋环境和模拟结果数据的交互应用。另一方面，目前的一些系统在设置相关模拟参数时需要较高的专业水平，使用友好性不足，静态的模拟结果也难以满足动态管理需求，求解方式的冗杂会引起系统总体效率的低迷。
针对上述问题，本研究对海洋环境数据结构，提出了海洋环境信息模型架构，可以实现多源异构海洋环境数据以及地形高程数据的自动获取、标准化处理、以及高效的组织管理，为后续研究提供了数据支持。进一步地，基于污染物本身的物化特性、扩散特征，从宏观和微观两个角度提出了适用于多尺度时空下海洋中各类典型污染物的二维扩散模拟分析模型，将污染物扩散过程分解为几个独立子过程进行分析，并通过不断对比优化确定了关键参数的自动化取值流程，在不明显降低模拟精度的情况下简化了计算量，节省了计算资源，缩短了模拟时间。在此基础上，对海洋环境标、矢量数据可视化技术以及污染物扩散模拟分析的交互应用技术进行了研究，基于 C#、python 和 WinForm 技术设计研发了海洋环境及污染物扩散可视化应用系统，实现了高效数据引擎、自动化信息服务以及污染物扩散模拟分析流程自动化，可以为污染物的动态追踪、预警、溯源提供相关技术与系统支持。
最后，对提出的模型及系统功能进行应用验证与分析。验证与分析结果表明，本研究提出的方法、模型和系统可以实现大尺度时空范围的海洋污染扩散模拟分析并且支持污染物微粒迁移路径追踪、指定位置污染物浓度变化分析以及污染范围计算等功能，并且与其他模型和分析平台相比，具有高自动化、高效率、低专业门槛的优势，可以满足海洋污染物防治、预警、动态追踪等实际需求，在海洋环境保护领域有重要研究意义与广阔应用前景。

The marine environment holds significant importance to human production and living. It is an essential component for achieving green, coordinated, and sustainable development. The simulation and analysis of pollutant diffusion in the ocean play a vital role in the management of coastal waters, marine pollution prevention, and early warning. Although there are currently many research and analysis systems related to this field, most of them are only suitable for simulating the diffusion of a single pollutant on a small spatiotemporal scale. Information barriers exist between these systems, making it difficult to achieve interactive applications of marine environment data and simulation results. On the other hand, setting relevant simulation parameters in some systems requires a high level of expertise, resulting in poor user-friendliness. Static simulation results are also insufficient for dynamic management needs, and overly complex solving algorithms may reduce system efficiency.
In response to the aforementioned issues, this study delves deeply into the structure of marine environmental data and proposes a marine environmental information model framework that enables the automatic acquisition, standardization, and efficient organization and management of multi-source heterogeneous marine environmental data and terrain elevation data. This provides data support for subsequent research. Furthermore, based on the physicochemical properties and diffusion characteristics of pollutants, we propose a two-dimensional diffusion simulation analysis model for various typical pollutants in the ocean under multi-scale spatiotemporal conditions, from both macro and micro perspectives. The pollutant diffusion process is decomposed into several independent sub-processes for analysis, and an automated parameter value selection process is determined through continuous comparison and optimization. This approach simplifies the computational load without significantly reducing the simulation accuracy, saving computational resources and shortening the simulation time. Building upon this foundation, we tackled the challenges of marine environment scalar and vector data visualization techniques and the interactive application of pollutant diffusion simulation analysis. Utilizing C#, Python, and WinForm technologies, we developed a marine environment and pollutant diffusion visualization application system. This system. This study has achieved systematic results in smart maturity assessment in the construction stage of engineering. Firstly, a general composition framework of smart maturity assessment systemsin the construction stage of engineering was proposed. Then, this framework was used to establish two specific smart maturity assessment systems. Finally, a representative industry was selected to practice smart maturity improvement strategies. In a word, this study provided an overall methodology and specific utilization instructions for the smart maturity assessment in the construction stage of engineering, especially construction management and floating wind turbine installation, which is of great significance and broad application prospects for the comprehensive development of smartness in the entire engineering field. achieves an efficient data engine, automated information services, and automation of the pollutant diffusion simulation analysis workflow. It can provide relevant technical and system support for dynamic tracking, early warning, and source tracing of pollutants.
Finally, we apply and analyze the proposed models and system functions. The results show that the methods, models, and systems proposed in this study can achieve large-scale spatiotemporal marine pollution diffusion simulation and analysis, support pollutant particle migration path tracking, designated location pollutant concentration change analysis, and pollution range calculation. Compared with other models and analysis platforms, it has the advantages of high automation, high efficiency, and low professional threshold. It can meet the practical needs of marine pollutant prevention and control, early warning, and dynamic tracking, holding significant research value and broad application prospects in the field of marine environmental protection.

大量先进技术和管理方法的应用促进了工程建设阶段的智慧化发展，从业组织因为巨大的竞争压力，都在积极探索建设阶段的智慧化转型及发展途径。然而，目前各工程领域普遍缺少针对从业组织自身或所负责项目在建设阶段的智慧化成熟度评价方法或体系，使得各组织无法准确掌握自身或具体项目在建设阶段的智慧化成熟度，也缺少合理可行的智慧化成熟度提升策略帮助它们制定准确并高效的智慧化发展计划。所以，智慧化成熟度评价方法的缺失极大地影响了工程建设阶段的智慧化发展进程。
针对上述问题，本研究通过文献调研、问卷调查、专家讨论和案例分析等方法，首先总结了用于评价工程领域建设阶段智慧化成熟度的体系组成框架，包括体系组成内容及具体构建流程，然后利用该框架分别建立了量化的针对建筑施工管理和浮式风机安装工程的智慧化成熟度评价体系。两套体系具备相同的结构，包括由评价维度和评价指标构成的智慧化成熟度评分表，作为评价结果分别从整体和具体角度呈现方式的智慧化成熟度等级和评价维度雷达图，以及智慧化成熟度提升策略。两套体系在初步建立之后都对其进行了验证与校准，通过案例分析证明了体系的准确性、合理性和实用性，两套体系都可以有效帮助各自行业内的从业组织在自评之后查漏补缺，并逐步提升自身或所负责项目的智慧化成熟度。最后，本研究参照提出的浮式风机安装工程智慧化成熟度评价体系，实际选择智慧化手段进行开发，建立了数字孪生框架，通过分析验证了该体系的积极作用。
本研究针对工程建设阶段智慧化成熟度的评价取得了系统性成果，首先总领性地提出了工程领域在建设阶段的智慧化成熟度评价体系组成框架，其次细分性地使用该框架建立了两套具体的智慧化成熟度评价体系，最后选取有代表性的行业实操智慧化成熟度提升策略，为工程领域尤其是建筑施工管理和浮式风机安装工程的智慧化成熟度评价与提升提供了总体的方法论和具体的使用说明书，对于整体工程领域的智慧化全面发展具有十分重要的意义和广阔的应用前景。

  The application of a large number of advanced technologies and managerial approaches has promoted smart development in the construction stage of engineering. Due to the tremendous competitive pressure, related organizations are actively exploring ways of smart transformation and development in the construction stage of engineering. However, there is currently a common lack of assessment methods or systems for the smart maturity of the organizations or projects they are responsible for in the construction stage of engineering, making it difficult for organizations to accurately grasp the smartmaturity of their own or specific projects in the construction stage of engineering, and there is also a lack of reasonable and feasible smart maturity improvement strategies to help them formulate accurate and efficient smart development plans. Therefore, the lack of smart maturity assessment methods has greatly affected the smart development processin the construction stage of engineering.
In response to the above issues, this study, through literature research, questionnaire survey, expert discussion, and case study, first summarized the composition frameworkof systems to assess the smart maturity in the construction stage of engineering, including the system components and specific construction processes. Then, two quantitative smart maturity assessment systems were established using this framework respectively for construction management and floating wind turbine installation. These two systems have identical structures, including smart maturity scoring tables composed of assessment dimensions and indicators, smart maturity levels and radar charts of assessment dimensions as the overall and specific presentation of assessment results, and smart maturity improvement strategies. Both systems were verified and calibrated after being initially established, and the accuracy, rationality and practicality of them were proven by case studies so that they can effectively help practitioners in their respective industries identify and fill gaps after self-assessment, and gradually improve the smart maturity of their own or responsible projects. Finally, this study selected smart measures for actual development referring to the proposed smart maturity assessment system for floating wind turbine installation projects so that a digital twin framework was established, andthe positive effect of this system was verified through analysis.

AEC（Architecture, Engineering & Construction，建筑、工程和施工）行业是典型的知识敏感型和风险敏感性行业，其在工程建设过程中高度依赖于国家规范、行业标准、集团内部规定等强制性文件，这给工程建设管理中包括项目管理条例编制、交底方案审查、施工报告验收等多个必要工作带来了较大的困难，因为这些工作所涉及的知识要点繁多且相互关联复杂，同时具体的合规性要求也因行业不断发展和规范不断细化而越发难以被人力所掌握。
另一发面，提高行业的 KM（Knowledge Management，知识管理）水平被认为有显著意义，而目前又已经有较多工作从领域规范性文本出发并展开研究，因为规范性文本包含了大量 AEC 相关的显性知识。因此，可以从这些方向出发，探索上述工程建设管理中典型问题的解决方案，提高工程建设管理的效率和质量。然而，目前相关研究仍然面临两个方面的不足：一是所处理的数据和验证的规模均较小，二是涉及的特征还较为单一。这两点都导致目前的相关研究面临泛用性和可靠性方面的问题。为处理上述问题，本研究首先收集了 307MB 的大规模规范性文本数据，并从多个粒度进行了复杂度降级和丰富的特征分析工作，以支持从规范性文本数据中批量化获取结构化信息。然后，本研究对工程建设管理的基本体系进行梳理，并以工程建设管理知识元对象视角作为切入点，生成了拥有 20 万节点和 200 万条边的大规模知识图谱，用于为相关知识要点及其语义关联的快速检索提供支撑。在此基础上，本研究进一步提出语义化的知识检索、评估、理解和应用全流程，并在工程建设安全管理条例的校核和高指向性的语义化安全知识问答两个方面实现具体的功能应用。最后，本研究对所提出的技术路线以及相应数据积累进行了平台化整合，完成了“特征→知识→应用→平台”完整流程。 
实例验证表明，本研究所构造的知识图谱总体质量较好，提出的工程建设安全管理条例的自动校核方法也获得 63%的 F1指标，已经可以在一定程度上支持相关的工作需求。总体上，本研究所提出的技术路线和相应数据积累提高了工程建设管理的自动化水平，并具有较大的应用前景。

  Architecture, Engineering & Construction (AEC) industry is typical knowledge & risk sensitive, and the process of engineering construction & management highly depends on mandatory documents such as national specifications, industry standards and internal regulations of the group, which brings great difficulties to many necessary tasks in engineering construction management, including project management regulations preparation, disclosure scheme review, construction report acceptance and so on. Because the knowledge involved in these tasks are more and more numerous and interrelated, and the specific compliance requirements are also more and more difficult to be mastered by human being due to the continuous development of AEC. 
On the other hand, it is considered significant to improve the Knowledge Management (KM) level of the AEC field, and there has been many works and researches based on AEC normaltive texts, as they contain a large amount of explicit knowledge related to AEC field. It is a feasible way to solve the typical problems above, and improve the efficiency and quality of engineering construction management. However, at present, the relevant research still faces two deficiencies: one is that the scale of data involved is very small, and the other one is that the characteristics involved are relatively simple. These two points lead to the lacks of universality and reliability in the current related research. 
In order to deal with the above problems, this study first collected a large scale of 307MB AEC specification text data. And then the hierarchy complexity degradation and feature analysis are carried out, to support the batch acquisition of structured information from the AEC text data. After that, this study combs the basic knowledge system of engineering construction management and defines the “meta knowledge object” of engineering construction management, based on which a large-scale knowledge graph with 200 thousand nodes and 2 million edges is built. The KG can be used to provide support for the fast retrieval of relevant knowledge objects and their semantic relationships. this study further discusses a complete KM flow with semantic knowledge retrieval, evaluation, understanding and application, with which two basic and valuable knowledge application is proposed: one is the engineering construction safety management regulations auto-complicance checking (ACC), and the other is the highly directional semantic safety knowledge question & answering (Q & A). Finally, the whole technical map and results and in this study is integrated to build the construction safety management KM platform, which realizes the complete flow of "feature → knowledge → application → platform". The case study shows that the overall quality of the KG constructed by this research is good, and the ACC of engineering construction safety management regulations also obtains 63% F1 score, which means it can be apply on relevant practical work to a certain extent. In general, the technical map and corresponding data accumulation in this study improve the automation level of engineering construction management, and have great application prospects. 

建风环境可对建筑人员、结构、设备与环境带来重大影响，是建筑全生命期需考虑的重要议题。目前，尽管以计算流体力学（Computational fluid dynamics, CFD）为代表的风环境分析技术已较为成熟，一系列建筑风环境分析工具也已被研发，但这些软件大多面向建筑设计阶段，基于当地常年气象统计资料与理想化边界条件进行风环境分析，具有静态性、理想化、求解效率低、专业门槛高等特征，难以满足建筑运维管理实时、实地、动态、易用的风环境分析需求。建筑运维阶段尚不存在适用、成熟的风环境分析应用解决方案，难以实现决策过程中对建筑风环境的定量、精确考虑，限制了建筑运维管理精细化、智能化水平的提升。
针对上述问题，本研究引入多尺度数值模拟与深度学习两项核心技术，对现有风环境分析方法与体系进行改进创新，构建基于多尺度建模与数据驱动的运维期建筑风环境分析框架，进而提出风环境数据管理、运维应用与平台研发方案，形成运维期建筑风环境分析的系统理论、技术、方法与平台。首先，调研当前运维管理与风环境分析流程特征，从数据、模型与应用等层面提出适用于运维期建筑风环境分析的理论体系与技术架构。面向建筑风环境分析应用所涉及的多源异构数据，提出统一信息模型，实现风环境分析应用全过程数据的自动获取、集成与管理。随后，构建耦合中尺度气象模式与微尺度 CFD 的多尺度数值分析模型，支持精确时空条件下的风环境模拟，满足运维阶段动态精确分析需求，并通过专家知识的嵌入实现模拟流程自动化。进而搭建面向建筑风环境预测的深度神经网络模型，以前馈式计算提升建筑风环境预测效率，并提出多尺度数值模拟-深度学习融合分析框架，实现风环境分析效率与准确性的兼顾。围绕污染物扩散、人员舒适性与建筑能耗三个关键运维议题，提出风环境信息在建筑运维阶段的应用途径。在上述关键技术的基础上，研发运维期建筑风环境分析应用平台，为考虑风环境因素的建筑运维决策提供完善的工具与平台支持，并结合实际建筑运维项目，对平台功能流程进行验证。
应用表明，本研究提出的理论、方法、技术与平台，可实现特定气象条件、精确时间范围、实际街区建筑群环境下的风环境动态分析，并支持分析过程的低专业门槛、自动化、高效率进行，满足运维期建筑风环境分析的实际需求，为考虑风环境因素的智能建筑运维提供了可行途径，具有重要研究意义与广阔应用前景。

The wind has a significant impact on building personnel, structures, facilities and the environment, and is an important issue to be considered throughout the building lifecycle. At present, although the wind environment analysis technology represented by Computational Fluid Dynamics (CFD) is relatively mature, and a series of analysis software has been developed, most of the software is developed for the building design stage. The current analysis process usually initiates wind environment analysis based on local meteorological statistics and idealized boundary conditions, which is generally static, idealized, inefficient, and has a high professional threshold. However, the building Operation and Maintenance (O&M) process requires management tasks to be real-time, on-site, dynamic, and easy-to-use. As a result, current wind environment analysis methods are difficult to meet the needs of the O&M tasks, and there is still no applicable and suitable wind analysis solution for the O&M process. It is difficult to realize the quantitative and accurate consideration of the building wind environment in the O&M decision-making process, which limits the improvement of the refinement and intelligence level of building O&M management.
To address the above problems, this study introduces two core technologies of multi-scale numerical simulation and deep learning to improve the existing wind environment analysis methodology, and constructs an original wind environment analysis framework towards the building O&M process based on multi-scale modeling and data-driven methods. Combining the technologies of wind environment data management, data application and platform development, this study proposes a comprehensive solution including theories, technologies, methods and platforms for building wind environment analysis in the O&M period. Firstly, this study summarizes the actual demand of wind information in the O&M phase and investigates the current wind environment analysis theory, and then constructs a theoretical framework for building wind environment analysis in the O&M stage from the aspects of data, models, and applications. To deal with the multi-source, heterogeneous and multi-scale data involved in the analysis process, this study proposes the Building Wind Environment Information Model (BWEIM) which realizes the automatic acquisition, integration and management of the related data,including meteorology data, wind data, buildings and geospatial data. Subsequently, a multi-scale numerical analysis model including the mesoscale meteorology model and the microscale CFD model is constructed to support the wind environment simulation under precise spatial-temporal conditions. The multi-scale model meets the dynamic and on-site analysis requirements in the O&M stage, and realizes the automation of the simulation process through the embedding of expert knowledge. A deep neural network-based wind environment prediction model is then developed to address the inefficiency problem of the numerical simulation model. And a framework that couples multiscale numerical simulation and deep learning methods is proposed to balance the efficiency and accuracy of wind environment analysis. Concerning three key issues in the O&M process including pollutant dispersion, personnel comfort and building energy consumption, this study proposes the application methods of wind environment information in the O&M decision-making process. At last, on the basis of the above-mentioned key technologies, this study develops the wind environment analysis and application platform for the O&M stage. The platform provides comprehensive tools for the O&M decision-making process considering wind factors. The feasibility and reliability of the platform are verified with actual building O&M projects.
It is proved that the theory, method, technology and platform proposed in this study can realize the dynamic, efficient, automatic and low-threshold analysis of wind environment under specific meteorological conditions, given time range, and actual urban morphologies. The proposed approach meets the actual needs of building wind environment analysis during the O&M process, and provides a feasible way for the wind-environment-involved intelligent O&M management, which has important researchsignificance and broad application prospects

建筑在运维阶段的能耗管理对于节省建筑的运营成本，优化建筑能耗和降低碳排放量都有着重要的意义，而数据驱动的能耗管理的精细化、综合化和智慧化，则是能耗管理未来的发展方向。我国的能耗管理已经迈出了数字化的第一步，但是当前的实践中仍然存在着诸多问题：建筑实体、人员和数字模型的三因协同问题缺乏系统的需求分析与理论架构，相关数据标准粒度粗糙、不够完善，不同来源的数据互相异构、难以融合，人员因素缺乏考虑、人机协同水平与交互方式相对有限等等，从而限制了能耗数据分析与应用，以及基于数据驱动的能耗管理的水平。因此，研究建筑能耗管理中建筑实体、人员和数字模型三个主要因素的角色与数据特点、互相作用机制以及具体的数据协同方法与技术至关重要。
本研究针对以上问题，以 BIM 和物联网技术为基础，综合运用模式识别、机器学习等新兴数据技术，对能耗管理中建筑实体、人员与数字模型等三因的协同模式、协同技术和集成应用平台进行了系统性研究。首先，本文对建筑能耗管理的内容和其中三因的角色和数据特点进行研究，提出了建筑能耗管理中三因的协同模式与架构，并对后续研究需要突破的关键技术进行了分析。针对能耗管理过程的数据需求，设计了标准化的静态、运行、环境、能耗和人员等信息的数据模型，为相关数据提供了标准化的描述方法和存储机制，并在此基础上研究了异构能耗数据的集成方法。对于人员信息的提取和检测，设计了基于集成学习的检测方法以解决标注数据缺乏和训练结果实时更新的问题；同时设计了基于墙面触控的与建筑的人机交互方法。最后，针对数据清洗与修复的应用场景，在协同数据的基础上，建立了通过机器学习方法进行数据异常检测和修补的方法，并通过比较预测结果对方法的效果进行了检验。
应用研究表明，本研究所形成的理论、技术、方法和系统，能够实现在建筑能耗管理中建筑实体、人员和数字模型三个因素的有机耦合，为建筑能耗管理实现精细化、智慧化和综合化探索了可行的技术路径，具有切实的理论意义和广阔的应用前景。

 The operation and maintenance management (O&M Management) of buildings bears significance in saving operational costs, optimizing energy consumption, and reducing carbon emission of buildings, and the future of data-driven building energy consumption management lies in delicacy, smartness, and comprehensiveness. Currently, China has taken the step forward into the digitization of energy consumption management, yet problems exist in its practices: relevant data standards are coarse in granularity and incomplete; data from different sources are heterogeneous and difficult to fuse; occupant factors are deficiently accounted, and the coordination levels and interaction methods between buildings and occupants are relatively backwards. Analyses and applications of relevant data are consequently also restricted. These problems confined the improvement of China’s building energy consumption management levels, therefore making it vital to research the roles, characteristics, interaction mechanisms, and specific coordinationtechniques for the three major factors in building energy consumption management, namely buildings, occupants, and digital models.
Addressing the aforementioned problems, and founded on BIM and IoT technologies, this study comprehensively employed emerging data technologies including pattern recognition and machine learning, and conducted systematic research on the coordinationarchitecture and techniques, and supporting platforms of the three-factor coordination in building energy consumption management. First, through studying the content of building energy consumption management, and the roles and data characteristics of the three factors therein, this study proposed a coordination model and architecture for the three factors, and further identified the key technologies to be researched in successive studies. Then, regarding data requirements in energy consumption management process, this study designed data models for information categories including static, operational, environmental, energy consumption, and occupants, providing standardized description methods and storage mechanisms for relevant data, and on this basis researched the integration methods for heterogeneous energy consumption data. Next, addressing the detection and extraction of occupant information, this study designed an ensemblelearning-based detection scheme to relieve the problems of insufficient labelled data and the synchronous updating of learning models, while also designed an interaction method between buildings and occupants based on wall-surface touch control. Finally, addressing the major application scenario of data cleansing, this study established algorithms for anomaly detections and fixtures on the foundation of coordinated data, and further validated the effectiveness of the algorithms through comparison of prediction accuracies.
Practical researches have suggested that the theories, techniques, methods, and the system established in this study suffice to achieve organic integration of the three factors of buildings, occupants, and digital models, thus blazed a technological path for realizing delicate, smart, and comprehensive management of the management of energy consumption of buildings, and therefore possess pragmatic theoretical significance, and broad prospects of application.

建筑设备运维管理期间会积累海量而复杂的数据文件，其中大量信息与机电系统相关，具体包括暖通，给排水和电气等专业设备的运行状态、能耗信息及维护维修知识等。这些信息大体上可分为两类，其一为时序性的监测信息，如机电设备运行状态，能耗信息等，其特点是会随时间增加而快速积累，不便于统计分析，目前主要的存储管理介质为电子表格；其二为非结构化的运维信息，如机电系统的设计与维护手册等，其特点是异源地分布在互联网中，对其全面整理较为困难，目前主要的存储管理介质为文档图片。由于当前运维信息存储管理方式不利于计算机进行高效的数据检索和知识学习，在实际运维管理中有多种情形，如能耗数据分析评估，应急管理中的问题构件定位等，依然需要相关专业人士基于多年的知识和经验积累进行决策分析，往往需要消耗较大的人力物力成本。
针对上述问题，本研究引入了较为前沿的语义网和知识图谱技术，并结合神经网络和深度学习模型，在建筑运维领域，分别针对时序信息和非结构化信息建立了监测信息本体和机电知识图谱。前者探究了空间本体，传感器本体及能耗KPI（Key Performance Index）本体的集成框架，给出了基于资源描述框架（RDF Schema, RDFS）的监测信息本体检索分析方法，并实现了基于KPI的能耗分析评估。后者探究了对机电知识的提取、管理与可视化描述方法，具体包括使用双向长短时间记忆（Bi-directional Long Short-Term Memory, Bi-LSTM）和条件随机场（Conditional Random Field, CRF）模型完成机电设备语料分析和实体发现，使用残差卷积神经网络完成机电知识的关系抽取。在此基础上，本文进一步讨论并给出了一种基于BIM（Building Information Model）和语义分析的机电系统逻辑关系自动补全方法。
案例应用与探究部分，本研究使用既有建筑2017-2018年的监测数据建立了监测信息本体，并对其年、月、日各颗粒度的相关能耗KPI进行了计算分析和数据可视化表达。此外还使用实际工程的模型数据对基于机电知识图谱的逻辑关系自动补全方法对进行了检验，取得了较为理想的结果。整体来说，本研究所建立的监测信息本体和机电知识图谱为本领域的相关研究提供了信息集成的技术框架，可在一定程度上满足建筑运维期信息管理的查询分析需求，有助于提高建筑运维期的信息综合管理和检索分析效率。

A large number of complex data files will be accumulated during the operation and maintenance phases, among which most information is related to MEP systems, including the operation status energy consumption information and maintenance knowledge of HVAC, plumbing and electrical equipment. This information can be roughly divided into two categories. One is the sequential monitoring information, such as the operation status and energy consumption information of MEP equipment, which is characterized by rapid accumulation over time and the main storage and management medium is electronic tables. The other one is unstructured operation information, such as the design and maintenance manual of mechanical and electrical systems, which is characterized by lack of computer readable knowledge, and distributed separatly in the Internet which means difficult to collate them comprehensively. The main storage management medium of them are document or pictures. Because the current method of operation and maintenance information storage is not conducive to efficient data retrieval and knowledge learning by computers, there are many use case in operation and maintenance phases, such as energy consumption data analysis and evaluation, locating component in emergency management, etc, are still needed to make decision analysis by engineer based on years of knowledge and experience accumulation, which always means requiring a large amount of manpower and material resources.
To solve the above problems, combining with neural network and deep learning model, the ontology of monitoring information and the knowledge graph of MEP have been constructed in this study. In the monitoring information ontology, the definition and integration of spatial ontology, sensor ontology, monitoring equipment and monitoring KPI are realized. The methods of sensor information parsing, transformation and retrieval based on RDF are given. AS for the establishment MEP knowledge graph, this study explores the key technologies related to deep learning, including MEP entity reorganization by Bi-LSTM model, MEP images classification and retrieval with VGG16 model, relationship extraction of MEP entities with residual neural network, and finally using neo4j to realize storage and visualization of MEP knowledge graph. With the logic information in knowledge graph, this paper proposes an approach to automatically generate the topological chain of MEP systems with topological analysis, which can improve the efficiency of facility management (FM) activities such as locating components and retrieving related maintenance information for prompt failure detection or emergency management.
In the case study section, the monitoring information ontology is developed by using the monitoring data of the existing buildings from 2017 to 2018. This paper calculates and analyses the annual, monthly and daily KPI of energy consumption. In addition, the prototype system of automatic completion of logic relationships was applied to a real-world project for validation. The results showed that the approach was able to generate topological chains of MEP systems with an average accuracy of over 80%. Overall, the monitoring information ontology and MEP knowledge graph developed in this study provide a useful framework for related research in this field, as well as contribute to the comprehensive information management and retrieval during O&M phase.

地铁不仅是便捷的交通纽带，也对沿线的城市规划具有积极影响和导向作用，地铁车站周边上马大批新的工程项目是十分常见的现象。这些周边外部项目开展相关土建活动时，不可避免地会对地铁结构的安全和运行产生影响。因此在外部项目施工前和各施工过程中，需要评估其对地铁结构的影响，保证车站和轨行区结构的安全，及时发现可能的安全隐患并加以控制。“地铁保护”正是专门对外部项目进行相关的隐患排查、监管、每日巡检、审批以及制定各种应急措施预案等工作的总称。地铁保护工作关注地铁既有结构与外部项目结构之间的相互影响，需要综合地铁结构、外部项目、周边地质层等诸多信息，结合相关规范确定外部项目对地铁既有结构的影响程度，具有工作参与方多、相关模型复杂、信息量大的特点，这些特点导致了当前“地铁保护”工作中存在的模型信息隔离、信息提取困难、数据处理分析手段有限等问题。
地铁保护中的困难，其根源是模型之间的“信息隔离”导致信息交互不便。要解决地铁保护中存在的“信息隔离”问题，BIM 技术是一个可行的解决方案。BIM（Building Information Model/Modeling，建筑信息模型）是面向建筑全生命周期，旨在解决建筑行业长期存在的“信息隔离”现象的新技术。本研究在调研地铁保护的需求后，提出了基于IFC（Industry Foundation Classes）的地铁保护信息模型（Metro Protection Information Model，MPIM），该信息模型针对地铁保护的领域特点，对IFC原始实体进行了扩展，定义多种共享层构件实体和对应的属性集、枚举类型，从而能够更为准确地描述、集成地铁保护中的各种信息。进一步以MPIM为数据源，针对地铁保护中的主要工作流程，提出了基于信息提取与分类的自动评审算法以及基于层次分析法和神经网络的巡检记录优先度评价算法，实现对地铁保护主要工作的计算机辅助和决策支持。
以课题组已有的BIM-FIM为基础，研发了基于BIM的地铁保护系统（BIM-MPS，BIM-based Metro Protection System），并应用于广州地铁八号线磨碟沙至新港东区间的地铁保护工作当中，以验证本研究提出的模型、算法的实际效果。经过试用，管理人员声称系统能够普遍节约评审时间40%以上。取得了良好的效果。也为进一步在地铁保护中应用BIM技术提供了借鉴和经验。

The metro is not only a convenient means of transport, but also has a guiding effect on the urban planning along the line. It is a common phenomenon that a large number of new construction projects are launched around the metro structure. When these surrounding external projects launched relevant civil construction activities, they will inevitably affect the safety and normal operation of the metro structure. Therefore, it is necessary to evaluate the influence on the subway structure before and during the construction of the external project to ensure the safety of the metro, detect and control the potential safety risks in time. ‘Metro protection’ is exactly the task of evaluating, detecting, and controlling the potential threats presented by external projects to metro structures, it’s in charge of the supervision, safety evaluation and regular patrols of the external projects and draws up detailed emergency preplans. Metro protection work concerns the interaction between existing metro structures and external project structures. To determine the influence level of external projects on the existing metro structure, various information must be integrated and comprehensive considered. Multi-participants, complexity of models and mass information are the three major characteristics of metro protection work. These characteristics have led to handicaps in the current metro protection work such as isolation of models, difficulties in information extraction, and the lack of data processing and analysis methods.
The “information isolation” is the root of various barriers in metro protection work, which leads to inconvenience in information exchange. BIM is a feasible solution to overcome the aforementioned isolation problem. BIM(Building Information Model/Modeling) is a new technology that aims to solve the long-standing “information isolation” phenomenon in the entire life cycle of architecture. After investigating the requirements of metro protection, this study proposed the Metro Protection Information Model (MPIM) based on Industry Foundation Classes (IFC), the general international standard for BIM information exchange. MPIM has expanded the original IFC by defining multiple shared-layer entities and corresponding property sets and enumerated types based on the domain features of the metro protection, thereby enabling more accurate information expression mode and integration of various information. Taking MPIM as a data source and aiming at the main work flow in metro protection, an automatic reviewing algorithm based on information extraction and classification, and an evaluation algorithm to quantitatively assess the priority and urgency based on analytic hierarchy process and neural network are proposed to realize computer-aided
evaluation and decision-making in the work process of metro protection. 
Based on the existing BIM-FIM of the research group, a BIM-based Metro Protection System (BIM-MPS) was developed. BIM-MPS was applied to the the metro protection work in the section between the Modiesha station and Xingangdong of the Guangzhou Metro Line 8 to verify the actual results and effects of the model and algorithm proposed in this study. Managers claim that the system saves more than 40% of the evaluation time and the safety scores given by the algorithm are rather satisfactory and reliable. Trial has achieved good results and also provided reference and experience for further application of BIM technology in metro protection.

疏散策略规划是大型公共建筑日常安全管理的基本任务。如今复杂而庞大的建筑物使这项工作变得更加困难。传统的疏散仿真通常需要对几何数据和属性信息进行单调乏味的建模，而疏散路径规划中不能直接使用设计数据。这会浪费人力和造成不准确的数据。另外，现有的基于图搜索的寻路算法在时间复杂度方面不尽如人意，并且难以处理实时情况。
为解决这些问题，本文提出了一种用于大型公共建筑物疏散规划的动态策略规划方法，核心算法基于BIM（Building Information Model）技术和各种机器学习技术。首先给出了一个疏散子模型的定义，该子模型能够表示所有需要的信息，包括疏散实体的定义和拓扑关系。还提出了应用机器学习技术自动评估区域安全等级的方法来完善疏散信息。根据施工阶段交付的BIM数据，本文进一步提出一种空间拓扑算法以自动建立空间拓扑结构。之后，将路径生成算法和局面评估方法相结合，生成了用于策略网络的巨大训练集。当主要策略网络得到充分的训练后，就将进入自学迭代的循环。然后提出一种动态多人策略算法来计算整个建筑物内的多人路径之间的交叉影响。这里策略网络可以将人员分散到更安全的地区，而且能降低平均的紧急度指标。最后提出数据后处理中的信息挖掘技术，并应用至实际项目的疏散模拟结果，以得到更深层次的有用信息。
在基于BIM的大型建筑安全管理平台BIM-SMP中编制了一个应用插件，称为基于BIM的动态疏散策略应用平台，作为验证本研究成果的原型系统。BIM可视化编程为功能点提供了形象的展示手段，实现了各种疏散元素的交互显示和路线的多种可视化效果。在案例研究中，生成了大约4000个公共建筑的有关疏散的实体信息。根据算法要求，共有八代策略网络接受了训练。现场BIM应用的两个典型使用案例已经验证了所提出的方法。后处理阶段还应用了混合数据挖掘技术。这些应用案例为大型公共建筑的安全管理提供了合理而有用的切入点，并且证明所提出的动态疏散策略方法可以在复杂、广阔的空间内组织实时紧急疏散，并有助于日常疏散管理以精确模拟各种情景。

Evacuation path planning is a basic task in routine safety management of large public buildings. With the development of functional and voluminous public buildings, safe and efficient evacuation under emergency scenarios remains a serious issue. Difficulties on evacuation path planning emerge in modeling methods, speed of algorithms, and reliability. Traditional evacuation simulation usually requires tedious modeling of geometry data and property information where design data cannot be directly used in the evacuation path planning. This brings waste of human labor and possibly inaccurate input. In addition, existing path finding algorithms based on graph searching can hardly handle changing real-time conditions.
In order to address these problems, this paper proposed a dynamic evacuation policy approach for evacuation path planning in large public buildings. The core method is based on Building Information Model and various machine learning technologies. First, a scheme of evacuation sub-model is given to represent all needed information, including evacuation entity definitions and topology relationships. Here an automated safety level evaluation based on two machine learning methods is proposed as an auxiliary algorithm. Space topology is automatically established according to delivered data from construction phase. Then, a path generation algorithm and an evaluating algorithm are combined to generate a huge training set for policy networks, where the analogy between evacuation paths and chess games is exploited to train the policy network. When the primary policy network is adequately trained, it falls into a self-learning iteration. A policy network is expected to output a correct path very quickly despite the changing environment but has not yet considered negative effects from other paths. A dynamic path planning algorithm is then proposed to calculate overlapping among paths all over the building. The policy networks can spread people out into safer regions, which has been verified in terms of reducing the average emergency level. Finally, some data mining skills are proposed towards evacuation simulation results for real building projects in order to find deeper information behind massive data.
An application plug-in was developed in BIM-SMP, a large-scale construction safety management platform based on BIM. It was called the BIM-based dynamic evacuation policy application platform as a prototype system to verify the research results. The BIM visual programming provides a flexible display method, enabling the interactive display of various evacuation elements and multiple visualization effects of routes. In the case study section, about 4,000 evacuation elements of a public building were generated according to the proposed scheme. Eight generations of policy networks were trained. Two typical use cases by on-site consultant groups have verified the proposed method. At post-processing stage, hybrid data mining methods are used in practice. All these applications provided reasonable and useful insights for safety management of large public buildings. The dynamic path planning method has the potential to organize real-time emergency evacuation in complex, wide space, as well as help routine evacuation management to simulate various scenarios with fine accuracy.

地铁建设作为城市基础建设的重要组成部分，已经成为城市发展的重要名片，甚至可以用来衡量一个城市的发达程度。我国主要城市的地铁建设水平依然落后于发达国家，在未来一段时间内，大力发展地铁建设依然是我国城市发展的主要旋律。
地铁施工作为地铁建设的最重要的一环，依然面临着很多问题。不同于一般的建筑施工，地铁施工中面临着诸多难题。主要体现在：临时占地的方案受周边环境制约难以确定，尤其是涉及的产权单位及属地管辖政府的要求难以满足；施工方案受周边环境影响难以确定；地下结构的复杂性与不确定性，使得施工过程中的风险很高，甚至于出现地面塌陷等严重事故；单位工程多，工序复杂，施工组织难度高；地铁工程往往是政绩工程，工期十分紧张；严格的环保政策，对于地铁施工长周期的土方作业的约束很大；地铁施工范围跨度大，安全质量隐患更容易出现。
本文以北京地铁8号线三期土建施工01合同段为例，在项目的前期准备阶段、临建建设阶段、初支暗挖阶段直到进入结构二衬阶段的全过程中，通过应用BIM技术，将过程中遇到的各类问题进行研究分析并解决，对施工方案进行优化设计，协助施工方案的制定。通过应用BIM技术，将各工序进行模拟建造，可视化交底，指导施工人员施工，发现工序中的关键信息，保证各工序施工组织合理，施工质量可控。利于相关的平台工具，摸索地铁施工中的信息化管理的模式。在技术人员及施工人员素质参差不齐的条件下，能够对其进行有效的培训指导。同时使用一种可以普及到施工最前线的信息传递方式，对变化的施工现场进行动态管理，及时发现各类隐患，及时处理，并将隐患信息分类分析，为施工现场的教育培训提供指导。应用一种基于二维码的管理技术，实现地铁施工中的设备维护的信息化管理及施工质量的可追溯性。
地铁施工中还是过多的依靠施工从业人员的经验和资历，这是行业的特性。通过本应用研究，为施工单位在地铁施工中如何应用BIM技术来协助从业人员解决施工中的问题提供了参考，并提供了一种简单便捷、经济有效的信息化管理方式，为BIM技术在地铁施工领域的推广提供了思路。

As an important part of urban infrastructure construction, subway construction has become an important business card for urban development and can even be used to measure the degree of development of a city. The level of subway construction in major cities in China still lags behind that of developed countries. In the coming period, vigorously developing subway construction is still the main melody of urban development in China.
As the most important part of subway construction, subway construction still faces many problems. Different from general construction, there are many difficulties in subway construction. Mainly reflected in: the temporary land occupation plan is difficult to determine by the surrounding environment constraints, especially the requirements of the property rights unit and the territorial jurisdiction of the government are difficult to meet; the construction plan is difficult to determine by the surrounding environment; the complexity and uncertainty of the underground structure, The risk in the construction process is very high, and even serious accidents such as ground collapse occur; there are many unit projects, complicated processes, and difficult construction organization; subway projects are often political achievements, and the construction period is very tight; strict environmental protection policies for the subway construction The earthwork of the cycle is very restrictive; the construction scope of the subway is large, and the safety quality hazard is more likely to occur.
This paper takes the contract section of the third phase of civil engineering construction of Beijing Metro Line 8 as an example. Through the application of BIM technology in the preliminary preparation stage, the temporary construction stage, and the initial branching stage of the project until the entry into the second lining stage of the structure, Research and analyze various problems encountered in the process, optimize the design of the construction plan, and assist in the formulation of the construction plan. Through the application of BIM technology, each process will be simulated and constructed, visualized to the bottom, the construction personnel will be guided to construct the key information in the process, and the construction organization of each process will be reasonable and the construction quality can be controlled. Conducive to relevant platform tools, explore the mode of information management in subway construction.
Under the condition that the quality of technicians and construction personnel is uneven, it can be effectively trained and guided. At the same time, using a kind of information transmission method that can be popularized at the forefront of construction, dynamic management of the changed construction site, timely detection of various hidden dangers, timely processing, and classification and analysis of hidden danger information, to provide guidance for education and training on the construction site. Applying a two-dimensional code-based management technology to realize the information management of equipment maintenance and the traceability of construction quality in subway construction.
In the construction of subways, it is still too much to rely on the experience and qualifications of construction practitioners, which is the characteristics of the industry. Through this application research, it provides a reference for how the construction unit applies BIM technology in the subway construction to assist the practitioners to solve the problems in the construction, and provides a simple, convenient and economical information management method. It provides ideas for the promotion of BIM technology in the field of subway construction.

建筑业一直存在信息化、智能化水平不高，管理效率低下等问题，在建筑的运营维护管理过程中尤其突出。建筑设备的正常运行是建筑内部良好环境的重要保障，因此建筑设备的管理是建筑运营维护管理中的关键任务。过去，建筑设备运行管理的信息化手段主要是应用建筑自动化系统（Building Automation System，BAS），但是，其较强的专业性导致运维管理人员难以有效使用。另外，设备运行监测数据仅仅存在于自动化系统中，产生“信息孤岛”局面，难以对监测数据有效利用。建筑信息模型（Building Information Model/Modeling，BIM）技术可以支持将设备监测数据与工程数据集成管理和应用，从而辅助运营维护过程中的综合分析与决策，以充分发挥工程信息的价值。
由于建筑自动化系统的传输协议不统一、监测数据量大、实时性要求高等特点，造成其与BIM系统集成过程中存在信息存储、集成和应用等多方面的问题。本研究首先从信息存储层面研究基于IFC（Industry Foundation Class，工业基础类）的信息扩展方法，利用扩展IFC属性集的方式实现在IFC中存储和表达建筑自动化系统中的监测信息。其次，从信息集成层面研究面向多协议的动态监测系统与BIM开放平台集成的关键技术，包括基于消息的分布式架构技术、基于插件的多协议扩展技术、时间序列数据处理技术、基于内存数据库的实时数据管理技术、基于时间序列数据库的历史数据管理技术和基于RESTful架构的监测数据查询服务技术。再者，从信息应用层面研究停车场监测信息在停车场智能管理中的应用、室内环境监测信息在舒适度评价中的应用、水位水质监测信息在水厂智能运维管理中的应用，扩展BIM的应用范围。最终，开发基于BIM的建筑设备自动化集成系统，并应用到合肥湖畔新城项目和北京槐房再生水厂项目。
应用表明，本研究提出的基于BIM的建筑自动化集成技术和所研发的系统能有效地解决监测数据与BIM模型的集成问题，可提高建筑运行维护管理水平，具有广阔的应用前景和价值。

The construction industry has always been at low levels in informationization, intelligentization and management efficiency, especially in the operation and maintenance (O&M) management. Since normal operations of construction equipment can ensure a good environment inside the buildings, management on the construction equipment serves as a critical task in O&M management. In the past, the informationization of O&M on construction equipment was implemented by the application of the Building Automation System (BAS). However, its demand on professionalism made the O&M personnel hard to effectively use the system. Besides, the equipment operation monitoring data only existed in the automation system, leading to a situation of “information island”, and the monitoring data cannot be effectively used. The Building Information Model/Modeling (BIM) technology can implement integrated management and application of monitoring data and engineering data, thus assisting in comprehensive analysis and decision-making during O&M to make full use of engineering information.
As BAS has inconsistent transport protocols, numerous monitoring data and requires strict real-time property, many problems occur in information storage, integration and applications, etc. amid integration with BIM system. This study firstly implemented storage and expression of monitoring information in the BAS by extending the Industry Foundation Class (IFC) attribute set using the IFC-based information extension method from the information storage layer. Secondly, it made a research on critical technologies for integration of the multi-protocol-oriented dynamic monitoring system and BIM open platform from the information integration layer, including the distributive architecture technology based on messages, multi-protocol extension technology based on plug-ins, data processing technology in time sequence, real-time data management technology based on memory databases, historic data management technology based on time sequence databases, and monitoring data query service technology based on the RESTful architecture. Thirdly, this thesis studied the application of parking monitoring information in intelligent management of the parking lot, the application of indoor environment monitoring information in comfort evaluation, and the application of water level and quality monitoring information in intelligent O&M of water plants to expand the application scope of the BIM. Finally, a BIM-based construction equipment automation integration system is developed and applied to the Hefei Hupanxincheng Project and Beijing Huaifang Water Recycling Plant Project.
The applications demonstrate that the BIM-based construction equipment automation system integration technology and the technologies proposed in this thesis can effectively integrate the monitoring data and BIM to improve the building O&M management, which has wide application prospects and values.

建筑业是世界上最大的行业劳动群体，也是国民经济的重要支柱产业。由于建设工程劳动密集、投资大等特点，导致建筑业安全事故严重，建筑施工安全问题严重影响了建筑业的可持续发展和社会稳定，成为建筑业发展的巨大障碍。
我国建筑施工安全水平较低，针对施工过程“时变结构”，缺乏先进、科学、实用的分析技术手段和管理工具，主要表现在：1）现有的基于静态结构的安全分析方法对建筑施工期的连续时变特性的描述和计算精度存在不足和局限，难以满足大型、复杂工程的施工安全分析；2）现有的基于时变理论的安全分析方法实际操作困难，用于整个施工期结构安全分析工作量巨大，自动化程度较低；3）现有的质量安全管控模式，更多的取决于足够数量的案例、数据和专家经验的支持，应用于施工现场的可操作性和准确性较差；4）由于导致施工安全事故的因素诸多、关系复杂，而且具有相当的不确定性和偶然性，难以实现对诸多因素的集成监控管理和综合分析管控。
为改变我国建筑安全技术落后的局面，引入信息技术是一条必需的途径。本研究通过引入建筑信息模型（（Building Information Model，BIM）和4D技术，将建筑结构安全分析与施工质量安全管控相结合，研究面向建筑施工过程新的安全性能分析和管控的理论、技术和管理工具，为实现自动化、集成化、动态化的安全分析与管控服务。首先建立一个面向建筑设计、结构分析及施工安全管控的扩展的4D施工安全信息模型（4DSIM++），并在此基础上研究基于4DSIM++的模型转换机制，实现建筑结构模型自动转换及多结构分析模型的联合对比；之后研究模型整合与数据集成技术，形成完备的4DSIM++并提出基于4DSIM++的安全分析与管控的整体应用流程以及一系列动态安全分析与管控的技术支持；最终研发具有自主知识产权的“基于BIM的时变结构安全性能分析与管控系统”（BIM-SAMS），并在实际工程项目中得以验证。
本研究实现施工期时变结构自动、连续、动态的安全分析以及集数据采集、动态集成、智能分析于一体的的质量安全管控，为施工期时变结构的安全分析与管控提供一套完整的理论、技术与平台支持，对于改变建筑施工安全管理手段落后的局面，提高安全水平具有重要的意义和广阔的应用前景。

随着我国建筑产业化进程不断推进，建筑工程也逐渐由现场施工转为工厂化预制加工的生产方式。尽管，对于后者而言，前期建立工厂以及购买设备的资金投入比较高，但实际工程数据表明，后者相比前者能够显著缩短工期、降低总成本、节省人力物力、降低环境污染等，因此近年来得到了越来越广泛的关注和应用。
其中，管道工程是当前建筑产业化中较为成熟的应用主体。其施工流程包括预制构件的深化设计、预制加工、运输和现场安装，并最终将设施实体竣工交付至运维管理。但此流程目前还存在如下几个问题：1）我国建筑管道工厂化施工方式仍然处于起步阶段，其工作流程及管理模式还不够完善；2）缺乏智能有效的深化设计工具，其深化设计过程严重依赖于设备厂家，设计过程不透明；3）信息传递以纸质或电子版文件为主，容易造成信息丢失与冗余，而且可能在多个地方都有备份，导致数据的不一致；4）缺乏信息共享机制，导致各阶段信息无法有效的从一个阶段传递到下一个阶段，形成“信息孤岛”问题；5）各参与方之间的沟通机制不完善，没有有效的协同工作平台，容易导致信息扭曲或丢失。
建筑信息模型（Building Information Model、BIM）是在三维数字模型的基础之上集成了建筑项目及产品的各种相关信息，形成的工程项目及过程的数字化模型。BIM技术可以有效的解决各阶段信息共享以及多参与方协同工作的问题，可以支持可视化、智能化的深化设计，从而为解决上述问题提供了技术支撑。基于这一理念和技术，本文面向建筑管道工厂化施工，首先研究了建筑管道工厂化全流程的技术和管理特点，接着对预制构件深化设计技术及多参与方业务流程管理这两个关键问题进行了深入研究，从而提出了基于BIM的建筑管道工厂化深化设计与辅助施工的解决方案。基于此解决方案，并结合课题组前期研究工作，设计并搭建了一个基于BIM的建筑管道工厂化信息管理平台（BIM-IMP）。该平台主要包含四个系统，分别为建筑管道工厂化预制构件设计系统（BIM-FDD）、建筑管道工厂化辅助施工管理系统（BIM-FC）、建筑管道工厂化移动平台信息管理系统（BIM-FM）、机电设备运维管理系统（BIM-FIM）。BIM-LMP可以解决建筑工厂化各阶段信息共享问题，为多参与方协同工作提供了统一的平台。该平台在两个大型项目中进行了应用验证，取得了良好的经济效益和社会效益，为进一步推动建筑工厂化和BIM技术的发展提供了良好的借鉴。

With the development of construction industrialization in China, construction engineering is also gradually transforming itself from onsite construction to factorial prefabrication process. Compared with the onsite construction mode, the factorial construction costs a lot to build factories and buy equipments in the early period. But it can also shorten the whole construction period, reduce total cost, save manpower and material resource, and is friendly to environment. Therefore it attracts more and more attention and has been applied to many projects.
Factorial pipeline construction is a mature part of current construction industrialization. The production process consists of several stages, i.e., detail design of prefabricated components, prefabrication processing, transportation, installation, and finally facilities delivery to operation and maintenance stage. However at present, this production process faces the following problems. 1) Factorial pipline construction is still in its infancy in China so that its working process and management method are inperfect. 2) With the lack of effective intelligent detail design tools, the detail design process heavily depends on the equipment manufacturers. 3) Information loss and redundancy, and data inconsistencies caused by paper-based or file-based information transfer. 4) Information isolation due to the lack of effective information sharing platform and methods. 5) Information distortion and loss result from an undeveloped communication between each stage and each participant.
Building Information Model (BIM) is a digital model of the project and process based on 3D digital model. It integrates various information related to construction projects and products. BIM can effectively solve the problems caused by information sharing and collaborative works in each stage and support the visualization, intelligent detail design. From the perspective of factorial pipeline construction, this research firstly studied the characteristics of the lifecycle process management of building pipeline factory. Then two key issues, i.e., detail design of prefabricated components and business process management for multi-participants were addressed, and corresponding methods were presented respectively. Based on these methods, a BIM-based solution to factorial construction oriented detail design and computer aided construction was porposed. Combining with some early achivements of our group, a BIM-based information management platform (BIM-IMP) for building pipeline factory was developed. On this platform, there were mainly four systems, i.e., a BIM-based detail design system for pipeline prefabricated components (BIM-FDD), a BIM-based computer aided construction management system (BIM-FC) for building pipelines, A BIM-based mobile system for factorial information management (BIM-FM) and a BIM-based intelligent facility management system (BIM-FIM). The BIM-IMP supported information sharing in each stage and provided a unified platform for the collaborative works among multi-participants. The platform was applied to two large-scale projects and achieved desirable economic and social benefits, providing a good reference for promoting building factory and BIM technology.

随着建设行业的互联网转型进程以及 BIM 技术在建设工程全生命期的持续应用，对三维可视化技术的要求逐渐提升，主要体现在：（1）需要 BIM 与 GIS 融合展示；（2）要求高渲染性能以支持高细度的大体量 BIM 三维数据；（3）拥抱以WebGL 为代表的互联网三维显示技术，实现跨设备、跨平台的全场景三维可视化。
针对建设工程项目施工阶段 BIM 与 GIS 在三维场景中融合展示的数据集成更新困难、渲染材质效果差、三维模型缺失定位基点、三维渲染性能低等问题，以开源的 Web 端 GIS 三维图形引擎 Cesium 为基础，以实际工程应用为出发点，研究建设工程三维模型数据接入方法，研究大规模三维模型在 GIS 场景中的集成与渲染方式，并以物联网与施工工艺模拟的可视化需求为例验证可行性。通过对接 Cesium 与 BIM 云平台，实现了 BIM 三维数据的动态集成与更新，为项目全生命期的三维场景展示奠定了基础。
通过解决 BIM 的地理定位问题，保证了 BIM 与 GIS 在 Cesium 中的集成展示。通过提出了相机移动期间高性能渲染以及基于堆的渲染命令限制方法，有效保障了 Cesium 的渲染性能，消除了 Cesium承载高 LOD 级别 BIM 模型的限制。在上述技术的基础上，针对物联网信息实时三维展示需求，实现了物联网设备与 Cesium 中三维模型的关联，在 Web 环境中实现了物联网消息订阅，并可实现 Cesium 三维模型实时更新。在射频识别的实时展示、梁场自动化喷淋以及人员室内实时定位等场景完成了具体应用。针对施工工艺三维动态可视化需求，提出了施工工艺可视化数据结构，研发了基于 Cesium 的三维动画快速生成方法，实现了施工工艺动画的快速更新与展示。这 2 个典型工程场景的应用验证证明了课题成果具备良好的可用性和应用潜力。

 With the gradual Internetization of the construction industry and the continuous application of BIM in the entire life cycle of construction projects, the requirements of 3D visualization technology have gradually increased: (1) BIM and GIS integration is required for display; (2) High rendering performance is required for supporting massive BIM 3D data with high precision; (3) Embrace WebGL to realize cross-device and cross-platform generic 3D visualization.
In order to solve the problems of difficult data integration and updating, poor rendering material effect, lack of positioning base point of 3D model, and low 3D rendering performance during the construction phase of BIM and GIS in the 3D scene, based on the open source Web3D GIS graphics engine Cesium, aimed to the actual engineering application, we study the data access method of 3D model of construction engineering, study the integration and rendering method of large-scale 3D model in GIS scene, and verify the feasibility by taking the visualization requirements of the Internet of Things and construction process simulation as an example.
By connecting Cesium and BIM cloud platform, we realize the dynamic integration and update of BIM 3D data, which lays a foundation for the display of 3D scenes in the whole life cycle of construction projects. By solving the geolocation problem of BIM, we integrate BIM and GIS in Cesium. By proposing high-performance rendering when camera moving and a heap-based rendering command limitation approach, we guarantee the rendering performance of Cesium and eliminate the limitation of Cesium to carry high LOD BIM data. On the basis of the above technologies, we realize the association between IoT devices and 3D models in Cesium for the real-time 3D display requirements of IoT information, realize IoT message subscription in the Web environment, and realize real-time updates of Cesium 3D models. Specific applications have been completed in scenarios such as real-time display of radio frequency identification, automatic spraying of beam fields, and real-time positioning of personnel indoors. Aiming at the requirement of 3D dynamic visualization of construction process, we proposed the data structure for construction process visualization, and a rapid generation method of 3D animation based on Cesium, which realizes the rapid update and display of construction process animation. The application verification of these two typical engineering scenarios proves that results of this study have good usability and application potential.

近年来我国建筑业得到了持续稳定的发展，建筑业总产值持续增长态势，但在当前管理模式下，建设工程的生产效率依旧底下，建筑业的生产效率反而有所下降智慧建造作为信息技术在建筑施工领域的具体应用已成为施工管理中新的热点，其实现离不开对数据的收集和处理。物联网技术作为实时数据采集的主要手段，在工程项目上建立物联网平台具有重要的意义。但设备协议多样，数据多源异构，数据量迅猛增长，无法集中管理，传统分析手段达不到时效性的需求，设备所采集的现场数据与 BIM、业务数据之间存在断层等问题阻碍着物联网技术在建设项目中的广泛应用。
针对以上问题，本课题综合应用物联网 BIM、云计算、大数据、人工智能等理论和技术，对建设工程物联网平台的体系架构、关键技术和支撑平台开展了系统性的研究。首先通过分析梳理建设工程物联网设备及其数据的类型和特点，提出建设工程物联网平台的技术架构，为关键技术的研究和平台的研发提供了理论基础和技术路径。针对终端设备多样、支持协议各异的问题，提出物联网网关的一般架构和三种具体的实施方式，实现了物联网设备的接入。针对物联网数据的特点，对上行和下行数据进行分析与分类，提出设备三级管理模型和上下行数据的标准格式，利用基于云的关系型数据库和时序数据库对物联网数据进行标准化管理和统一存储，并实现了物联网数据与 BIM 数据的融合。针对具体业务场景，研发了规则引擎和摄像头物体识别和定位的技术方案，支持设备报警、规则联动、工人穿着识别、工序识别、物体定位等具体的智能业务应用。在关键技术突破的基础上，设计研发了建设工程智能物联网平台，为建设工程施工阶段的物联网数据采集、转换、存储、流转和智能应用的完整流程提供支撑平台和工具。最后，结合实际工程项目，对课题所提出的方法、技术和平台进行了应用验证。
应用表明，本研究所形成的理论、技术、方法和平台，实现了建设工程施工过程中物联网数据从采集、传输、存储、分析到智能业务各环节的应用，为智慧建造中的物联网平台应用探索了可行的技术途径，具有较高的理论意义和广阔的应用前景

 China's construction industry has been developed in recent years, and the total construction industry output value continues to grow trend, but in the current management mode, the production efficiency of construction projects remains underneath, the production efficiency of the construction industry has declined instead Intelligent construction as a specific application of information technology in the field of building construction has become a new hot spot in construction management, its realization cannot be separated from the collection and processing of data. IOT technology as the main means of real-time data collection, the establishment of IOT platform on engineering projects is of great significance. However, problems such as diverse equipment protocols, heterogeneous data from multiple sources, rapid growth of data volume, inability to centralize management, failure of traditional analysis means to meet the demand of timeliness, and disconnection between field data collected by equipment and BIM and business data hinder the wide application of IOT technology in construction projects.
In view of the above problems, this topic integrates the application of IOT BIM, cloud computing, big data, artificial intelligence and other theories and technologies, the construction engineering IOT platform system architecture, key technologies and support platform to carry out a systematic study. Firstly, by analyzing and sorting out the types and characteristics of construction engineering IoT devices and their data, the technical architecture of construction engineering IoT platform is proposed, which provides the theoretical basis and technical path for the research of key technologies and the development of the platform. For the problem of diverse terminal devices and different support protocols, the general architecture of IoT gateway and three specific implementation methods are proposed to realize the access of IoT devices. For the characteristics of IoT data, the upstream and downstream data are analyzed and classified, the three-level management model of devices and the standard format of upstream and downstream data are proposed, the cloud-based relational database and the time-series database are used to standardize the management and unified storage of IoT data, and the integration of IoT data and BIM data is realized. For specific business scenarios, the technical solutions of rule engine and camera object recognition and positioning are developed to support specific intelligent business applications such as equipment alarm, rule linkage, worker wearing recognition, process recognition, object positioning, etc. Based on the key technology breakthrough, the construction engineering intelligent IOT platform is designed and developed to provide a support platform and tools for the complete process of IOT data collection, conversion, storage, flow and intelligent application in the construction stage of the construction project. Finally, the methods, technologies and platforms proposed in the subject are applied and verified in conjunction with actual engineering projects.
The application shows that the theory, technology, method and platform developed in this research realize the application of IOT data from collection, transmission, storage, analysis to intelligent business in the process of construction engineering construction, exploring a feasible technical way for the application of IOT platform in smart construction, with high theoretical significance and broad application prospect.

在现有的土压力研究基础上，本文将其扩展到考虑地表超载影响的有限宽度黏性土土压力研究中。以弹塑性理论为基础，针对富水软土地区密集地下空间环境中，邻近大刚度既有地下结构的深基坑工程，考虑了有限黏性土土体宽度、墙土间相互作用和地表超载的影响，分析周边有限土体的受力特性，提出基坑墙后有限土体及其宽度的明确定义，建立了一种修正的土压力理论计算模型。进而合理确定有限土体滑裂面形态，得到基坑墙后有限土体土压力变化基本规律，并采用数值模拟、深圳地区工程现场实测和离心机模型试验的方法，对理论研究成果进行验证。
结果表明，与现有土压力理论相比，本文方法与现场实测、数值模拟及离心机模型试验结果吻合较好，说明本研究成果对有限宽度黏性土土压力理论改进效果显著。有限土体滑裂面服从双折线模型，穿过挡土墙墙趾，并与既有地下结构侧壁相交，直至沿侧壁到达地表。基坑与既有地下结构间有限土体的土拱效应明显，基于土拱效应理论，将土体小主应力轨迹线改进为左右对称的悬链线模型，其力学机理与实际情况更相符。通过改进的有限宽度黏性土土拱理论，主要解决了三个问题：1）墙土之间的相互作用可通过墙土间外摩擦角综合表示；2）明确了有限土体的临界宽度范围；3）确定了土压力随着有限宽度黏性土物理力学性质的变化规律。进而得到土压力及其系数分布、墙顶张拉裂缝埋深、土压力合力及其作用点位置的改进理论模型。有限土体土压力为非线性分布，随着有限土体宽度的增加而增大，增大趋势减小。若忽略土体黏聚力和有限土体宽度的影响，将使得基坑围护结构的整体稳定性大幅下降，大大增加事故发生的风险。当基坑土压力及其作用点距墙趾的距离较大时，由此设计得到的基坑支护体系抗倾覆稳定性较好。

 This paper extends the existing earth pressure research results to the earth pressure research of limited-width cohesive soil. Based on the elastoplastic theory, for the deep foundation pit adjacent to the the existing large-rigidity underground structure among the intensive underground space in the water-rich and soft soil area, take into account the influence of the limited-width cohesive soil, the interaction between the wall and soil, and the surface overload, analyze the stress characteristics of the surrounding limited soil, then clearly define the limited soil and soil limited-width behind the foundation pit wall to establish a theoretical calculation model of modified earth pressure. Reasonably determine the slip surface shape of limited soil to obtain the basic change laws of the limited soil earth pressure behind the foundation pit retaining wall. The numerical simulation, field measurement of SHENZHEN projects and centrifuge model tests are used to verify the theoretical research results.
The results show that the theoretical results of this paper are in good agreement with the results of field measurements, numerical simulations and centrifuge model tests compared with the existing earth pressure theory, indicating that the theoretical results of this paper have a significant improvement on the theory of earth pressure of limited-width cohesive soil. The slip surface of limited soil is a double polyline, which passing through the retaining wall toe and intersecting the side wall of the existing underground structure until it reaches the surface along the side wall. The soil arching effect is more obvious in the limited-width soil between the foundation pit and the existing underground structure. Based on the theory of soil arching effect, the arch curve model has been changed into a symmetrical catenary model and the mechanical mechanism is more consistent with actual. The improved theory of soil arching effect of limited-width soil mainly answers the following three aspects, (a) The interaction between wall and soil can be comprehensively expressed by the external friction angle between the wall and soil, (b) Clarified the critical width range of limited soil, (c)Determine the change law of earth pressure with the physical and mechanical properties of limited-width cohesive soil. Obtain the theoretical model of earth pressure and its coefficient distribution, the depth of the wall tension crack, the resultant force of earth pressure and its action point. The limited-width soil earth pressure is nonlinear distributed and increases with the increase of the limited soil width, and the increasing trend decreases. Ignoring the influence of soil cohesion and limited soil width would greatly reduce the overall stability of the foundation pit retaining structure and increase the risk of accidents. The longer distance between the action point of the earth pressure on the foundation pit and the wall toe is better in terms of stability of support system against overturning.

风灾造成的损失中，建筑风灾致损占据主要部分。因此，做好建筑风灾的灾前评估预防和灾后应对对于防灾减灾工作而言相当重要。现今已有多种策略和方法可用以研究建筑在风环境中的承受的作用及响应，以试验、计算方法为主流。这些方法往往结果精确，可迁移性强；然而各方面的成本较高，且需要一定的专业基础。因此难以适应对于更广泛、更快速的建筑风环境灾害灾前评估的需求。频发的风灾为研究提供了大量的数据，采用适当的技术手段，可以从中发掘出建筑风灾致损的相关信息和规律。据此，本课题通过收集既往风灾数据，使用机器学习、最近邻检索、知识图谱等技术方法，实现建筑风灾致损情况预测、相似案例检索、知识自动问答等功能，构建建筑风环境灾害评估平台，实现对建筑在一定风环境下可能发生的破坏的快速预测判断，并能提供相关知识，为制定风灾应对策略提供参考。

 As one of the most significant natural disasters in China, losses resulted from wind disaster is costly, among which the damage of buildings accounts for the main part. Hence, it is important to do well in forecasting, evaluating and responding to the damage of buildings in wind disasters for disaster prevention and mitigation. Nowadays, there are many strategies and methods to study the effect and response of buildings in the wind environment, with experimental and computing methods as the mainstream. These methods are often accurate and transferable accompanied with high cost in many aspects and requirements for professional foundation. Therefore, it is difficult to satisfy the needs for a wider and faster assessment of building wind environmental damages. Frequent wind disasters provide a large amount of data for research, where relevant information and laws can be excavated with appropriate techniques. Accordingly, this paper builds a building wind environmental disaster assessment platform by collecting previous data, with functions as damage prediction, similar case recommendation, automatic question answering achieved, utilizing machine learning, nearest neighbor search, knowledge graph along with other technical methods. It serves for providing rapid prediction and assessment of the possible damage of buildings in a certain wind environment, and offering relevant knowledge as reference to.  

公路建设工程是一类复杂的、安全风险较高的系统工程，具有工类繁多、施工流动性大、环境多变等特点，存在较大的安全风险。传统的、粗放的管理模式难以支撑有效的安全管理，公路工程领域亟待探索以知识支持为导向的精细化、高效化管理模式。目前，公路工程行业已积累了大量安全生产资料可供行业人员参考，但对于如何从来源各异、价值密度低的海量数据中快速、精准地查询到所需知识，尚缺少行之有效的办法。基于此，本研究立足于公路工程安全管理领域，以数据驱动为导向，利用知识提取的方法构建了公路工程安全领域的大型知识图谱和领域专业术语库，并提供基于知识图谱的智能化知识支持解决对策，具体工作包括：构建公路工程安全管理领域的知识建模体系，从特征、过程、管理、组织、对象、要素等多维度进行知识分解，形成面向全种类、全生命周期扩展的多维度的公路工程安全领域层状知识体系，作为知识提取的本体层结构。实现面向公路工程安全数据的知识提取过程。基于采集到的公路工程安全领域的海量数据，设计以自然语言处理为基础的知识提取算法流程，完成领域知识的自动化提取。结合关系类型的扩展和规则的提取，构建领域专业术语库，引导算法流程的局部优化，以提升知识提取结果的准确度。对生成的知识结构进行实体、关系、数据的融合，形成大型知识图谱，作为知识提取的数据层。面向应用场景，对知识图谱结构进行加工和重组。分析知识图谱在实际应用场景中所需的结构组织形式，对知识图谱本体层和数据层进行关联，并以术语库为重要依托，实现数据层向本体层的自动化扩展。进而基于本体层和数据层之间的映射，进行知识结构的局部加工和重新组织，以支持实际场景中的应用需求。

 Highway construction is a complex and high-risk systematic engineering with quantities of work branches, high mobility and changeable environment, which brings about great safety risks. Since the traditional and extensive management mode is difficult to support effective safety management, it is urgent to explore a refined and efficient management mode guided by knowledge support in the realm of highway engineering. At present, a large number of safety production data has been accumulated in the highway engineering industry for the personnel to refer to, yet in practical application, there still exists a deficiency of effective solutions to get the quick and accurate access to the required knowledge from the massive data with different sources and low value density. Therefore, based on the field of highway engineering safety management and driven by data, this research aims at applying knowledge extraction method to build a large-scale knowledge map and a professional term database within the domain, and providing intellectual knowledge support solutions based on knowledge graph.Specific contents include: Build the knowledge modeling system related to highway engineering safety management, and decompose the knowledge from the multi-dimensional aspects of characteristics, process, management, organization, objects and elements. Further, produce a multi-dimensional layered knowledge system in the realm of highway engineering safety oriented to the expansion of full category and full life cycle. It is used as the ontology layer for knowledge extraction. Realize the knowledge extraction process for highway engineering safety data. Based on the collected massive data, an automated knowledge extraction procedure based on natural language processing is designed to accomplish the extraction of domain knowledge. Combined with the expansion of relationship types and the extraction of textual rules, a professional term database is constructed to guide the local optimization of algorithm procedure, so as to improve the accuracy of knowledge extraction results. The generated knowledge structure is further fused with entities, relationships and data to form a large-scale knowledge map, which is used as the data layer for knowledge extraction. Oriented to application scenarios, the knowledge map structure is further II processed and reorganized. This research analyzes the knowledge structure and organization format required in the actual application scenario, associates the ontology layer and the data layer of the knowledge map, and realizes the automatic expansion from the data layer to the ontology layer based on the important support of term database. Then, relying on the mapping relationships between ontology layer and data layer, the knowledge structure is locally processed and reorganized to support the application requirements in the actual. 

为了解决我国乡村住宅设计与建造中由于信息化程度低下而导致的一系列问题，我们提出建立一个面向乡村住宅的基于Web的GIS与BIM轻量化平台。目前，平台的集成工作与轻量化工作尚未完成。本文基于FBX SDK提出了一种文件的自动集成方法，该方法具有自动化与可拓展的特点。本文提出了一种基于顶点重要度的网格简化算法，该算法用顶点重要度与模型特征因子控制模型的简化过程，在简化过程中保护乡村住宅模型的轮廓与细节特征。这两种技术扩充了平台的使用场景，提高了平台的模型处理能力，有利于平台的推广使用。

In order to solve the problems caused by the low level of informatization in the de-sign and construction of rural housing in China, we propose to set up a lightweight plat-form of GIS and BIM based on Web for rural housing. At present, data integration and lightweight has not been completed. Based on FBX SDK, this paper puts forward an integration method for FBX files, which is automatic and extensible. This paper puts forward a lightweight algorithm for rural residential information model. Vertex im-portance and model feature factor are used to control the process of mesh simplification, so that the contour and details of the model can be better protected. These two technolo-gies expand the applicable scenarios of the platform, improve its model processing ability, and are conducive to the dissemination of the platform.

随着我国建筑行业的高速发展，建筑领域在各个方面积累了大量的原始数据，如何从这些琐碎的数据中获取结构化的知识成为了行业智能化所面临的新问题。本文提出了从万维网上获取建筑领域文本数据的两种方式，获取了大量的文本数据。在实体提取和关系抽取过程中，我们分别应用和评估了BiLSTM-CRF模型、ResCNN模型，结果表明它们具有良好的性能。此外，我们设计并实现了具有预标注功能的实体标注工具，以期解决建筑领域标注数据缺乏的问题。

With the rapid development of the AEC industry in China, a large volume of original data have been accumulated in the industry. How to obtain structured knowledge from these trivial data has become a new problem faced by the intelligentization of the industry. In this paper, we proposed two methods of obtaining text data of the field of AEC from the Internet, and in turn extracted a large volume of text data. The BiLSTM-CRF model and ResCNN model are respectively applied and evaluated in the process of entity extraction and relationship extraction, and the results demonstrated good performance. In addition, we designed and implemented an entity annotation tool with pre-annotation function, in order to solve the problem of the shortage of annotated data in the field of AEC.

建筑MEP领域是建筑工程中的重要内容，是建筑在运维期得以正常工作的重要基础。得益于我国在近十几年中建筑行业的高速发展，MEP领域在各个项目档案、技术规范、学术文献、社区讨论中都积累了丰富的资料，若能利用其中的数据和信息，那么对于MEP在设计和运维阶段的自动化实现都能有相当大的帮助。同时，随着NLP技术和知识工程在近年来的探索与成熟，我们希望能够通过NLP相关技术，抽取MEP领域文档信息，以帮助我们更好的推进 MEP 领域智能化实现。在本文中，我们将针对MEP领域语料库，使用LSTM进行命名实体识别，然后使用CNN进行关系抽取。这两步都将基于远程监督和bootstrap的思想来尽可能的提高标注效率和预测质量，同时这两步得到的成果将整理为三元组构成知识图谱中的数据层。另外，在本文中将对这一数据层使用neo4j以及基于Django的web服务进行可视化展示。

The field of building MEP(mechanism, electrical, pumping) is an important part of building engineering, as well as an important basis for building to work normally during operation and maintenance. Thanks to the rapid development of the construction industry in China in recent decades, the field of MEP has accumulated abundant information in various project archives, technical specifications, academic papers and community discussions. If the data and information can be used, it will be of great help to the automatic implementation of MEP in the design and operation stage. At the same time, with the exploration and maturity of NLP technology and knowledge engineering in recent years, we hope that we can extract information from MEP domain document through NLP related technology, so that it can help us better promote the Intelligent Implementation of MEP domain. In this paper, we will use LSTM for named entity recognition in MEP domain text corpus, and then use CNN for relation extraction. These two steps will be based on the idea of “remote supervise”and “bootstrap” to improve the labeling efficiency and prediction quality as much as possible. At the same time, the results of these two steps will be sorted out into triples to form the data layer in the knowledge graph. At the same time, this data layer will be visualized using neo4j and Django-based web services.

随着城市对地下空间的开发利用，盾构机作隧道结构施工的大型综合性设备应用日趋广泛。盾构机在施工过程中，将记录大量监测数据，这些数据反映了盾构机的工作状态，对施工质量、成本与安全具有重要意义，但由于其数据量大、影响因素多等特点，目前利用率仍然较低。针对上述问题，本文将数据挖掘技术应用于盾构机监测数据中，对盾构机数据实时挖掘与分析技术进行深入研究。首先，本文分析了盾构机的系统组成以及监测数据的物理意义，随后分别实现了三类数据挖掘技术的应用。借助关联规则发现算法，本文对盾构机参数间的关系进行挖掘，并对挖掘结果进行解释说明，给出关联规则代表的的实际施工规律。借助分类算法，本文由盾构机监测数据构造决策树模型，实现了对盾构机所处地层的实时预测，并构建三维地质模型对挖掘结果进行可视化展示。借助 BP 神经网络，本文实现了对盾构机掘进速度的预测，通过预测值与实际数据对比，进一步给出了可能存在异常的施工工段。经实际数据验证，本文所采用的算法预测精度高、输出质量好，计算时间可满足实时分析要求。数据挖掘结果可为盾构施工提供信息参考，规避隧道施工风险，提高盾构机掘进效率。

With the development of underground space in cities, the use of tunnel boring machines(TBM) as large-scale comprehensive equipment for tunnel construction has become increasingly widespread. During the construction process, a large number of
monitoring data will be recorded, which reflects the working status of TBM and is of great significance to construction quality, cost and safety. However, the current data utilization rate is still low due to its abundance of records and influencing factors. To solve the problems above, a data mining method is applied to TBM monitoring data and an in-depth study on real-time TBM data mining and analysis is conducted. First, this paper analyzes the system composition of the TBM and the physical meaning of its monitoring data, and then realizes the application of three types of data mining techniques. With the use of association rule discovery algorithm, the relationship among TBM parameters is discovered and the construction law is given by the explanation of mining results. Classification algorithm helps construct a decision tree model which realizes the real-time prediction of stratum on which TBM is located, and the classification results is visualized by a three-dimensional geological model. Through the introduction of BP neural network, this paper realizes the prediction of tunneling speed of TBM. And the construction section with possibility of abnormality is further given by comparing the predicted value with the actual data. The case study shows that the algorithm used in this paper has high prediction accuracy, good output quality and short
calculation time which can meet the requirements of real-time analysis. The results of data mining can provide information for shield construction, avoid construction risks, and improve tunneling efficiency.

随着我国公路化建设的不断深入，行车荷载、施工维修和自然侵蚀等因素所造成的路面损伤在道路使用过程中产生的负面影响越来越不可忽视。及时地检测和识别这些路面损伤对道路的正常使用和长久使用便具有重要的意义。卷积神经网络等图像识别技术的成熟，提供了一个可以利用机器自动识别替代人工识别的机会。
本课题将利用图像识别技术，研究其在路面损伤识别领域的应用。通过研究学习基于卷积神经网络的图像识别技术的理论支撑和实现路径，探究各类平台和深度学习框架的特征与优劣，总结归纳研发背景和用户环境等需求分析，设计相应的路面损伤识别实现流程和程序代码，解决该领域在实际工程的应用问题。
本研究根据上述技术方案，开发了基于图像识别的路面损伤识别系统，并在实际工程中完成了检测和改进。应用表明，使用图像识别技术进行路面损伤识别的应用开发能够满足对主要路面损伤种类进行识别的需求，同时也支持在其他图像识别领域的迁移，是一个值得深入研究的技术领域。

With the continuous developing of China’s highway construction, the negative impacts of road damage caused by factors such as vehicle load, construction maintenance, and natural erosion are increasingly becoming more and more important. The timely detection and identification of these road damages is of great significance for the normal and long-term use of the road. The maturation of image recognition technologies such as convolutional neural networks provides an opportunity to use automatically recognizing machines to replace artificial recognition.
This topic will use image recognition technology to study its application in the field of pavement damage recognition. Through researching and learning the theoretical support and implementation path of image recognition technology based on convolutional neural network, the characteristics and advantages and disadvantages of various platforms and deep learning frameworks are explored, and needs analysis of the R&D background and user environment are summarized, and the corresponding pavement damage identification is designed. Realize the process and program code, and solve the application problem of the field in the actual project.
Based on the above technical scheme, this study developed a pavement damage identification system based on image recognition, and completed detection and improvement in actual projects.
 The application shows that the use of image recognition technology for road damage identification application development can basically meet the needs of the classification of the main road damage, but also supports the migration in other image recognition areas, is a very promising, worthy of in-depth exploration of the technology.

随着我国社会经济的发展和科学技术的进步，对于智能建筑的需求也开始逐渐上升。传统的智能建筑解决方案，虽然在一定程度上实现了建筑系统的自动化，但是通常是基于一套预设的判断条件，对变化的建筑环境和人缺乏适应能力。近年来快速发展的发展的人工智能技术为解决这一问题带来了转机。本课题针对建筑运维系统中的子系统照明系统，根据信息物理系统的架构，设计和提出了一套智能照明系统解决方案。本文首先提出了基于 Arduino 和树莓派等低智能设备，使用多种传感器实现信息采集系统，同时利用基于 ZigBee协议的智能照明设备作为实现器。然后，本文提出了基于被动学习策略的改进的聚类分析算法，来进行用户的模式识别，并使系统对用户行为具有适应性；同时提出了改进感知哈希算法和 SIFT 算法，进行图像距离计算和用户行为识别，保证了系统运行的准确性和流畅性。本文还基于 HTTP 协议设计了网络系统。之后，本研究对系统进行了实现并作了实例测试，系统稳定性、准确性和适应性良好。本研究的解决方案能够推广到其他运维系统，提高运维智能化程度，并在工业智能运维产品中加以应用。

With the development of China’s economy and advancements in science and technology, the demand for intelligent buildings is also on the rise. However, traditional solutions to building intelligence rely on pre-set conditions to perform, therefore although automate maintenance and operation systems to some extent, lag behind in adjusting to changing environments and occupants. Artificial intelligence technology, which underwent dramatic progress in recent years, shed light on this problem. This paper focuses the lighting system, a subsystem of the maintenance and operation system of buildings, and designs an intelligent lighting system solution based on cyber-physical system architecture. This paper first proposes utilizing sensors, based on low-intelligence devices such as Arduino microcontroller and Raspberry Pi to assembly the information collecting system, and utilizing pre-manufactured programmable lights,
which are based on ZigBee protocol, to assembly actuators. Then this paper proposes an improved clustering algorithm based on passive learning strategy, to perform pattern identification on occupants, while enabling the system to adjust to behaviors of occupants. This paper proposes along improved perceptual hash algorithm and improved SIFT algorithm to calculate similarity of pictures and identify occupants’ actions. This paper also assemblies networking system using HTTP protocols. The proposed design is implemented and a test is carried out, which came out to show the robustness, accuracy and adjusting ability are all acceptable. The result of this research can be applied in other operation and maintenance subsystems and improve its intelligence level, while also carries a prospect of industrial application.

在再生水厂的运行过程中，可能发生来水量骤增等不利情况，如果对再生水位各区域的水位调控不当，内部机电设备将会被水淹没，进而影响整个水厂的运转，造成重大损失。本研究针对以上问题，基于IFC（Industry Foundation Classes，工业基础类）标准和监测数据，建立再生水厂水位动态模型，并应用到再生水厂水位调控过程中，增强再生水厂运维过程中应对危机的能力。首先，通过对IFC标准的信息存储结构进行研究，利用属性及扩展的方式对IFC进行扩展，实现水位和水量监测信息在IFC中的表达，为再生水厂动态模型的建立奠定基础。然后，并对再生水厂内的监测系统组成进行调研，并对BIM（Building Information Model/Modeling，建筑信息模型/建模） 与水厂监测数据集成的方法进行了研究。接着，基于人工神经网络算法实现再生水厂来水量预测，建立计算模型实现水厂各区域水位的动态模拟预测和分析。最后，基于课题组已有的BIM-FIM平台开发基于BIM的水厂水位控制系统，并将其应用到槐房再生水厂的运维管理过程中。应用结果表明，该系统能够辅助水厂管理人员进行水区信息和控制节点信息的可视化查询、水位分析预测等，提高运维效率。

In the process of the operation in the recycling water plant, there may be some unfavorable conditions such as sudden increase of the amount of water. If the water level is improperly controlled, the electromechanical equipment will be flooded, and correspondingly, the operation of the whole water plant will be influenced. Based on the IFC standard and monitoring data, I established the dynamic model of the water level and applied it to operation period of the water level to enhance its ability of handling with unpredictable accidents. Firstly, I studied data structure of IFC standard, and extended it to achieve the expression of water level and monitoring data. Then, I learned about the composition of the monitoring system in the recycling water plant and the method of integrating the BIM with monitoring data. Next, based on the artificial neural network, I established a forecasting model, and used it to predict and analysis the water level. Finally, I developed the BIM-based water level control system based on the BIM-FIM platform and applied it to the operation and management of the Huaifang Recycling Water Plant. The application results showed that the system could assist the manager to carry out the visualization of water information and control node information, analysis and forecast of water level, and improvement on the efficiency of operation and maintenance.

目前对于BIM在运维管理中的应用是建筑全生命期各阶段中的热点研究领域。针对竣工交付工程中的信息整合与传递，目前虽然对于交付内容和技术有一定研究基础，但缺乏适用于现场交付的移动端App等相关辅助工具。。而要自主开发基于BIM 的运维App，一个较为重要的问题便是需要解决移动终端不同操作系统的兼容性。本文将利用跨平台架构，研究基于BIM-FIM扩展的竣工交付移动应用，通过对跨平台开发技术的比较分析和深入探讨Phonegap的实现原理，提出基于BIM的跨平台移动架构。然后通过对移动运维关键技术的研究，包括二维码扫描与构件识别、移动交付模型的信息集成与关联、移动端图形平台开发等，给出相关功能的实现方法。本研究根据上述技术方案，开发了基于BIM的跨平台竣工交付App，并进行了应用测试。应用表明，使用跨平台技术进行BIM移动应用开发基本能够满足BIM相关功能的需求，同时也支持相关功能的扩展，是一个十分有潜力，值得深入探索的架构。

The BIM-based facility management system that is currently applied in a broad range of applications and research. This study focuses on the intergration and delivery of completion model. Although there is a certain basis for the content and technology of delivery of completion model，we still need some mobile Apps and other related tools for the onsite delivery. As for developing BIM-based App, One of the difficulties is mobile application programs today require frequent cross-equipment compatibility. Using cross-platform development technology, this study developed a mobile application on the basis of BIM-FIM. First, based on the comparative analysis of cross platform development technology and the realization principle of Phonegap, this paper proposed a reasonable BIM-based FM cross-platform framework. Then, this paper discussed some key technologies of mobile FM and proposed implement methods of them, including the QR code scanning and component identification, the integration and connection of information for completion BIM model and the development of mobile 3D graphic platform. The BIM-based mobile App for FM was implemented and tested. The application shows that the use of cross-platform technology in BIM mobile application development can satisfy the needs of partial functions of BIM application and supports the extension of relevant functions. This framework has high potential and is worth further exploring. I will use this as the basis to further explore the cross-platform development of mobile applications in the BIM-based FM.

建筑工程运维阶段具有时间跨度大、涉及人员多的特点，在信息的获取和利用方面具有诸多困难，BIM的提出与应用为建筑工程运维管理带来了集成化的变革，但仍存在着研究不充分、数据利用困难的不足。本文针对运维期的大量数据集合的高效信息获取的需求，首先分析了BIM数据库的存储逻辑结构，构造了数据仓库。然后对运维阶段可能用到的三种信息挖掘技术做了深入研究和应用。在聚类分析中，以大型建筑在运营维护期产生的大量设备维护维修数据为算例，进行了完整的聚类过程。然后讨论了四种聚类情况，给出在运维管理中的应用方法。在频繁模式挖掘中，以Apriori算法为基础，提出一种空间复杂度和一种时间复杂度的改进算法。在离群点分析中，针对检测BIM中文件关联错误的需求，对关联构件的属性和文件名施行离群点检测过程。本研究还提出一种衡量检测离群点结果准确度的方法。本研究所采用的算法效率高，输出结果可解读性良好。研究成果可以指导进行维修决策、预测和人员材料的部署，作为预警的根据，亦可辅助决策，做到人力资源的高效利用。促进运维阶段信息的利用效率。

Operation and maintenance (O&M) period of buildings involves wide time range and numerous stakeholders, which cause significant difficulty in data retrieval and application. Introducing BIM into O&M has brought integration revolution though a lack of study and trouble in data usage still exists. This paper is aimed at the need in sufficient information retrieval from large data sets during O&M. First, logical storage structure of the BIM database was analyzed, and data warehouse was constructed. Then, three potential information mining approaches in O&M were deeply studied and implemented. In cluster analysis, a complete clustering process was carried out through a calculation example of mass repair and maintenance data set during O&M in large buildings. Then four kinds of clustering result and application methods were discussed. In frequent pattern mining, one improved algorithm focused on spatial complexity and one focused on time complexity were proposed based on Apriori algorithm. In outlier analysis, towards a need in detecting match errors between BIM and files, an outlier detection was carried out in element properties and file names. Further, this paper proposed a method to evaluate the accuracy after detection. The algorithms introduced in this paper were all effective with fine comprehensibility. Study result can be a guide in repair decision, prediction and early warning, as well as assist decision making leading to efficient resource usage and high efficiency during O&M period.

建筑信息模型（BIM）凭借其信息共享能力和可视性等优势，在当前建筑全生命周期管理中起到了革命性的作用。将BIM技术应用于建筑运维管理阶段仍处于起步阶段，但国内外都在进行积极的尝试，BIM技术在运维管理中发展前景光明。本研究选取运维管理的一个具体方面——停车场管理进行深入的调研和研究，并尝试将BIM技术应用在停车场管理领域。目前商用停车场管理系统很多，但尚未有将BIM应用于该领域的先例。考虑到BIM技术的优势，以及当前停车场系统功能上的欠缺，因此运用BIM技术进行停车场管理，弥补当前停车场系统功能上的不足，并发挥BIM的独特优势进行功能创新很有发展前景和研究价值。因此，本研究提出了一套基于BIM的智能停车场管理技术方案，以清华大学BIM-FIM平台为基础进行二次开发。首先在BIM-FIM中建立停车场BIM模型，设计数据库，完善停车场BIM模型的数据，并在BIM-FIM实现对停车场信息的查看、修改、导入等操作。同时，集成现有的车牌识别技术，获取实时停车信息和数据在BIM-FIM上反馈给管理者。最后，针对目前小区侵占私人车位问题，在BIM-FIM平台上实现车位分配，并设计算法实现车位引导。
本研究根据上述功能设计和技术方案，开发了基于BIM的智能停车场管理系统，并在清华大学何善衡楼进行实地应用测试。测试结果表明该系统的信息显示、修改，车位状态实时检测，车位分配和引导均可正常运行和使用。该系统对将BIM技术应用于运维管理进行了尝试，并取得良好效果，表明BIM在建筑运维管理中有良好应用前景。

Recently, Building Information Modeling(BIM) has been playing a vital role in building lifecycle management with its unique advantages in sharing information and visibility. Although the application of BIM in Facility Management is still in early stages, researchers in China as well as researchers abroad are making active attempts in it. To sum up, it has a bright future to apply BIM to Facility Management and other related areas.
This study conducts an intensive and comprehensive study on car park management, which is a specific field of Facility Management, and tries to apply BIM to the field of park management. There are many commercial park management system at presents, however, none of them integrate park management with BIM. Considering that BIM has unique edge in sharing information and visibility, it’s promising and worth studying to use BIM in park management to improve current park management system and create innovative functions in the field of park management.
Therefore, this study proposes intelligent park management solutions based on BIM, which uses BIM-FIM as its basis for secondary development. Firstly, the solution builds a 3D model of car park and develops databases for the modeling of car park. After that, the solution designs a set of windows forms for users to view, edit and add the data related to car park. Meanwhile, the automatic license plate recognition module is appended to the park management solutions, which can obtain the real time data and send them back to park managers. Lastly, to cope with a growing problem that private parking spaces are frequently occupied without owners’ permission, park managers can allocate empty parking spaces to visitors using this intelligent park management system. To bring more convenience to the park managers and drivers, the system will automatically provide guidance to the drivers to access their parking spaces.
This study completes the intelligent park management system based on the functional design and technical solution above. To test the functions and efficiency of the system, this study conducted an experiment in front of Heshanheng Building in Tsinghua University, and the result shows that all the modules function properly. This intelligent park management system based on BIM made attempts in applying BIM to Facility Management and achieved good results, indicating that BIM has a bright prospect in Facility Management and is worthwhile to do more researches in this field.

目前，国内BIM的发展如火如荼，BIM的合理运用能为建筑的规划、设计、施工和运维带来诸多好处。但是，BIM的应用主要集中在设计和施工阶段，而在建筑全生命期总成本控制的重要阶段——运维阶段的应用较少。在建筑运维管理中，机电设备的管理是非常重要的一部分，利用BIM对机电设备进行管理具有很大的前景，但是国内这方面的研究并不多，尤其在对机电设备BIM模型异构数据（运行状态的监测信息和监测数据等）的扩展和基于BIM的机电设备运行状态的监测分析等方面的研究还比较匮乏。
本研究提出了一套基于BIM的机电设备运维期监测技术方案。首先，在对建筑机电设备的组成和监测内容分析的基础上，提出基于传感器和建筑设备自动化系统的机电设备监测技术方案。然后，重点研究BIM模型监测信息的扩展，包括监测点信息和监测数据的导入接口、数据库层面监测信息的存储和访问机制、监测信息和BIM模型的关联机制等。最后，在清华大学BIM-FIM平台的基础上实现BIM模型监测信息的显示。
本研究根据上述技术方案，开发了基于BIM的机电设备运维期监测原型系统，并进行了应用测试。应用表明，该系统能够基本实现监测点信息和监测数据的导入和管理，在BIM模型中基本实现监测信息显示、历史监测信息查询、监测信息管理和机电信息查询等。该系统针对目前BIM研究比较匮乏同时又有很大价值的问题，具有较大的应用前景。

At present, the domestic development of Building Information Modeling（BIM）is in its full swing. The proper use of BIM can bring a lot of benefits to the planning, design, construction and operation and maintenance of the building. However, most BIM applications mainly focus on the design and construction phaseinstead of the operation and maintenance phase, which is an important stage for the cost control of buildings' life-cycle. In buildings' operation and maintenance management, the management of mechanical and electrical equipment is a very significant section. Though the utilization of BIM for mechanical and electrical equipment management has promising future, the amount of domestic researchs in this area is not so large, especially in the area of the integration of monitoring data with building information model, and the BIM-based the monitoring and analysis of the operating conditions of mechanical and electrical equipments.
This studyput forward a set of plans on monitoring techniques of mechanical equipments' operating and maintainance process based on BIM. First, based on the composition and construction of electromechanical equipment to monitor the content analysis of the proposed scheme based on electromechanical equipment monitoring sensors and building automation systems. Then, Ifocused on the integration of monitoring data with building information model, including the import interface of the monitoring point information and the monitoring data, the storage and access mechanism of the monitoring information in database level, and the approach to integrate the monitoring information with building information model. Finally, a prototype has been implemented to display monitoring information in 3D platform. The prototype is based on the BIM-FIM, which is developed within the department of Civil Engineeing at Tsinghua University.
The BIM-based prototype system for the monitoring of electrical and mechanical equipment was implemented and tested. The application shows that the system achieved the basic import and management ofthe monitoring data, and basically realizedthe monitoring information displaying within BIM model, monitoring information management, historical monitoring information queries and electrical and mechanical information inquiries. The system has great value and application prospect.

建筑信息模型（BIM）是数字技术在建筑业中的直接表达，它的提出对于整个建筑行业的信息化及其应用具有划时代的意义。现如今，BIM技术的运用对于解决建筑模型与结构模型转换、多种结构分析模型转换、结构分析模型信息全面共享以及建筑全生命期应用起着至关重要的作用。
本研究通过对结构分析模型转换机制的广泛调研学习，定义基于BIM的通用模型数据标准，通过对ETABS、SAP2000、ANSYS等结构分析软件进行研究，撰写其解析大纲，并运用C#语言编写各分析软件与BIM之间的双向接口，从而建立结构分析模型转换系统，最终实现任意两种分析模型之间的相互转换。
该转换系统提供模型显示、选择、缩放等功能，可以实现分析模型之间几何信息、截面信息、材料信息、荷载信息的成功转换，不仅可以摆脱之前繁琐的、重复性的建模工作，节省时间，提高质量，同时也为后续的结构优化提供可靠的建筑结构模型信息，具有较高的应用价值。

Building information modeling（BIM）is the direct product of digital technology, which has an epoch-making significance on the construction industry informatization and its application. Nowadays, the use of BIM technology plays a vital role in solving the conversion between building model and structure model, the conversion and comprehensive information sharing among the structural analysis models and the application through the whole lifecycle.By comprehensively researching on the transformation mechanism of the structural analysis model, we define a common BIM-based data standard. Then we write the outlines about some structural analysis softwares such as ETABS, SAP2000, ANSYS and so on. Moreover, we use C# language to build bidirectional interface between the above softwores and BIM, so as to establish the structural analysis model conversion system achieving the conversion between any two analysis modelsThe system can provide the functions of display, select, zoom of the model and achieve the conversion among the analysis models about the information of geography, section, material and load. It can not only get rid of the tedious and repetitive modeling work so as to save time and improve quality, but also provide reliable architectural and structural information model for the subsequent structural optimization. Therefore it is of high value in practice.

BIM（Building Information Model，建筑信息模型）技术是目前建筑行业的一种非常先进的技术管理手段。在建筑行业中应用BIM，可以有效地提高项目各参与方之间进行信息交换的准确性和效率，并有助于制订合理的施工方案，精确掌握项目进程，最终达到合理控制成本，缩短工期，提高施工质量的目的。但是目前市场上的BIM相关软件一般要求使用者具有较强的专业知识背景，对于非专业人员则很难对软件进行操作。因此现在BIM软件的使用者还主要集中于一些专业的公司和研究机构。另外，由于BIM是面向建筑全生命周期的，因此所涉及的信息量非常庞大，BIM的这一特点也会给BIM软件的使用带来麻烦。
本文所涉及的研究内容主要是，参考中国目前所制订的标准和规范，针对新型BIM平台进行了IFD（International Framework for Dictionaries，国际字典框架）数据库的整理和归纳工作，整理成树形结构，并提出了一种自动进行IFD词汇库整理的机制；利用NLPIR（Natural Language Processing and Information Rtrieval，自然语言处理与信息检索）分词工具，对用户所输入的内容进行语义分析；完成了BIM相关信息在数据库中的存储；最后初步建立了BIM信息查询系统。
本课题的目标是，通过进行相关研究，建立一个新型的BIM信息搜索平台，让那些不具备复杂专业知识的一般用户，利用本平台也可以方便地对BIM软件中的海量信息进行提取和利用，促进BIM技术的推广和普及。

BIM服务器的诞生为BIM对建设全生命周期的支持提供了解决方案。虽然目前BIM服务器的主要结构为C/S架构，但随着移动平台的逐渐普及、移动技术的逐渐发展，B/S架构与移动平台的良好兼容性这一优势也渐渐凸显。同时，伴随着HTML5标准的发布，我们迎来的不仅仅是支持浏览器硬件图形加速的WebGL技术，更有一系列划时代的Web技术，它们是将来的内嵌式富互联网应用的基础。本课题旨于实现B/S架构下的BIM信息管理与模型三维浏览。
本研究通过分析BIM的模型信息范围与表达形式，创建了适合于本系统Web客户端加载与信息显示的数据接口。同时，研究了基于WebGL框架Three.js的BIM模型显示技术，并据此实现了图形平台的开发。另外，还研究了面向Web的BIM信息管理技术，实现了数据传输效率的提升以及及时信息交互功能。基于以上研究成果，最终形成了较为完整的BIM模型三维浏览与信息显管理系统。
本研究对该BIM信息系统进行了应用测试。应用表明，系统的图形平台性能稳定，渲染效果较好，能够支持较大的BIM模型；而系统为Web客户端提供了丰富的功能支持，更体现出良好的交互特性。与类似的项目成果对比能够发现，该系统具有更强的针对性和更多的功能，因而具有较大的价值与应用前景。

The birth of the BIM Server provides a solution for BIM to support construction life cycle. The main structure of the BIM Server is C/S, but the advantage that the good compatibility of B/S with the mobile platform highlights gradually with the increasing popularity of mobile platforms and evolving of mobile technology. Meanwhile, with the release of the HTML5 standard, what we ushered is not only the WebGL which supports hardware graphics acceleration in browser, but also a series of epochal Web technologies which are the base of the future embedded rich Internet applications. This study aims to achieve the BIM information management and 3D viewing in B/S environment.This study created a data interface which is suitable for the Web client of our system to load and display information. Meanwhile, the BIM display technology based on Three.js which is a WebGL framework has been studied, and then, the graphical platform was developed by using it. In addition, the Web-oriented BIM information management technology was also studied, which realized the improvement of data transmission efficiency and timely information interaction. Based on the above research, a relatively complete system which is called BIM 3D display and information management system is finally formed.In this study, the BIM information system was applied and tested. The application shows that the graphic platform of the system has stable performance and nice rendering ability, and can support comparatively large BIM models. The system also provides rich function supports for the Web client, which shows good interactive features. Comparison with the similar projects, we can find that the system is more targeted and has more functions, which makes it has greater value and application prospect.

如何对运维期的桥梁结构进行高效的评估和维护维修管理一直是桥梁管养部门和市政桥梁管理部门十分关心的问题。随着计算机的发展，桥梁管理系统应运而生。然而，当前的桥梁管理系统普遍存在着管理体系不完整、分析和评估模型落后且不准确、数据库功能不够强大、人机交互体验性不强等问题。针对这些问题，本文以研发一个广义的桥梁管理系统为目标，提出了基于桥梁全生命周期的管理方法和流程，建立了科学的桥梁分析和评估模型，给出了基于B/S架构和WebGIS地图服务的桥梁管理系统的核心技术方案。该系统具有完善的数据库管理功能，强大的模拟分析能力，科学的决策支持以及友好的人机交互界面。由于该软件工程的规模较大，系统的复杂程度高，需要分阶段开展研发工作。在现阶段，本文主要做了如下几方面的研究：
1.	简单回顾了桥梁管理系统的发展历史及现状，并介绍了桥梁管理系统的基本概念和应用，总结了国内外桥梁管理系统实践的主要经验。
2.	分析了美国Pontis系统和中国的CBMS系统的功能模块组成和系统的功能缺陷，着重对CBMS系统的问题进行了详细的剖析，总结了桥梁管理系统开发的最佳实践方法。
3.	论述了桥梁管理系统的分析与设计全过程，提出了桥梁全生命周期管理策略，并给出了符合实际的管理流程，对系统中使用的结构退化模型与费用模型进行了详细的介绍。
4.	详细讨论了基于B/S架构的桥梁管理系统的核心技术方案，并对开发过程中的技术难点进行了总结，介绍了目前系统的开发进度以及已经实现的主要功能。
5.	结合实际，对未来桥梁管理系统的发展提出了思考和展望

住宅建设项目具有量大面广，标准化程度高，利润空间小，生产周期短的特点，因而，对住宅项目策划进行精确地进度、经济、绿色性能评价与分析至关重要。BIM（Building Information Modol，建筑信息模型）技术则可为住宅项目策划的各类分析与评价提供了完整、准确地模型和数据支持，从而完好地解决绿色住宅方案比选与决策的难题。Autodesk Revit是当前国内外广泛使用的BIM建模软件，然而，由于方案策划阶段BIM建模主要关注住宅产品定位、建筑面积、楼盘价格等宏观参数，Revit并不适用于方案策划阶段的建模。
本课题通过研究基于BIM的户型拼装、数据库、绿色建筑分析评价以及Revit二次开发技术，针对住宅项目方案策划，提出了户型拼装方法和基本的户型采光分析算法，建立了以户型拼装形成标准层、标准层复制建立建筑单体、多个建筑单体形成小区规划的BIM建模整体方案和方法，并设计了户型库。在此基础上，开发了基于Revit的绿色住宅参数化建模系统，实现了由户型到标准层，标准层到建筑单体，建筑单体到小区规划的“一条龙”建模，户型采光分析及其自动化参数导出等功能。
本研究开发的基于Revit的绿色住宅参数化建模系统应用于中建地产住宅项目方案建模，经实际应用，取得了良好的效果。验证了本研究提出的户型拼装方法和采光分析算法的有效性，户型拼装形成标准层、标准层复制建立建筑单体、多个建筑单体构成小区规划的“一条龙”建模方法的可行性。实际应用证明，本研究提出的建模方法及开发的基于Revit的绿色住宅参数建模系统，对于住宅方案策划具有较好的应用价值。

本项目为"十二五"项目的基础探索研究，结合知识引擎强大的知识管理分析能力，有效地对BIM模型中的海量数据进行分析，为用户提供良好、直观的分析结果，为项目决策、实施提供数据支持。
项目首要任务是建立我国的建筑术语库，即中文形式的IFD，建立自然语言与BIM模型中数据信息的对应关系，为知识引擎的信息提取与分析奠定基础；
其次，结合建筑全生命期某一阶段的实际应用需求建立相应的语义分析与推理规则，为自动根据语义进行数据分析计算提供逻辑支撑。

结构设计需要应用多种有限元软件进行联合分析和对比，然而各种结构分析软件之间的模型信息难以相互兼容。尽管目前不乏研究人员针对各自的需要，开发了针对某两种或某几种特定结构分析软件之间的数据接口，但其转换能力各异且局限性较大，难以推广。
因此，为搭建结构设计分析软件之间信息流和工作流沟通的桥梁，本研究旨在引入BIM的理念及其设计和实现的思路，面向结构设计阶段建立一个统一的信息模型，用于统一存储和管理结构设计过程中产生的所有信息，并实现其共享和集成。基于此模型，通过对多种结构分析软件（ANSYS、ABAQUS、MARC、ETABS、SAP2000、SATWE、MIDAS等）进行数据解析，统一各种软件的结构分析模型，实现各软件与BIM之间的双向数据接口。在此基础上，研发一个基于Web和BIM的结构分析模型转换系统，实现以BIM为核心的多种结构分析模型之间的自由转换和综合管理。
本研究为模型间的有效转换提供一套完整的数据解析及理论、技术、方法和工具，可以使结构工程师摆脱繁琐、重复的建模工作，节省时间、提高质量。