摘  要

       地铁工程建设是一项复杂的高风险性建设工程。一方而，国内地铁工程建设规模逐年扩大，另一方面地铁及地下工程施工安全事故频发。总结地铁工程目前安全管理的现状，即有规模大、发展快，技术和管理力量难以保证等客观原因，更重要的是对风险辨识不细致全面、风险控制措施不到位等主观原因。在地铁车站施工风险识别所需的经验丰富的技术人员十分匮乏地情况下，如何将专家经验和规范中的风险识别知识结构化，通过计算机自动识别风险将具有较大的理论研究和工程实践价值。

       本文提出研发基于工程图纸的地铁车站施工安全风险自动识别系统(SafetyRisk Identification System for Metro Station Construction on the Basis ofConstruction Drawings, SRIS)，通过研究风险致险因素自动提取方法，从标准、规范及专家经验中获取风险识别规则，设计风险自动识别推理机制和算法，实现由工程图纸信息到施工安全风险的自动输出。SRIS系统应用于地铁车站施工准备期风险评估过程，排查重大安全隐患，为风险动态控制与预警提供参考依据，是信息技术改革传统的风险识别过程的具体应用。

       设计并实现风险致险因素自动提取。首先对地铁施工安全风险及致险因素进行分类，然后建立众工程图纸中可识别的项目特征类、结构构件类、地质水文类和施工环境类信息与各类致险因素的关联关系表，最后设计计算机识图策略和专用算法图纸中各类信息的识别和工程语义解析，从而完成工程技术参数及致险因素的自动提取。

构建基于可信度的风险识别知识库。总结地铁车站施工安全风险识别领域的相关理论性知识和专家经验性知识，系统地归纳整理地铁车站施工安全风险及其致险因素，阐释了风险识别知识获取过程、风险识别知识的混合表示方法及结构化存储的方法，在此基础土构建基于可信度的风险识别知识库。

  提出基于规则推理（RBR)和基于CBR-RBR混合模式两种推理机制。针对风险识别规则的构成特点，在基于规则推理(RBR)原理基础上提出基于元规则控制的成组检索匹配推理算法和风险可信度计算方法。总结安个事故案例的多层次属性特征，提出基于案例推理(CBR)和基于规则推理(RBR)混合模式在地铁车站施工安个事故案例库中的应用模型，并指出在CBR主导前提下RBR的具体应用方法，实现从施工安全事故案例库获取风险识别信息。

  研发并应用基于工程图纸的风险自动识别系统。分析和设计地铁车站施工安全风险自动识别系统的总体结构、功能模块划分、数据库结构。以武汉地铁二号线工程螃蟹甲车站项目为例，给出了风险自动识别的全过程和实例验证。

关键词: 地铁车站    施工安全     风险自动识别    工程图纸      致险因素

Abstract

 The metroproject is a high risk construction project. On the one hand the overall constructionscale increases in China year after year, On the other hand metro and undergroundproject construction safety accident occured frequently. Summarizing the currentstatus of safety management in metro construction, there are objective reasons suchas huge investment scale, fast development pace, lackage of technical and managementstrength, the subjective reasons that risk identification is not thorough, riskcontrol measures is not in place are more important. How to make riskidentification knowledge in the experts experience and the regulations struturedand identify the metro station construction risk automaticall by computer is atopic of greater theoretical research and engineering practice value in thesituation of very scarceful of experienced technical personnel.

The primaryobjective of this paper is to propose Safety Risk Identification System forMetro Station Construction on the Basis of Construction Drawings (SRIS).Through the study on the risk factors automatic extraction method，risk identification rules can be acquiredfrom the standard, specification and expert experience, inference mechanism andalgorithm of risk automatic identification

can be designed，realizing safety risks in metro stationconstruction can be output automatically by the engineering drawinginformation. SRIS can identify potential

safety hazard for risk dynamic controland warning in the construction preparation period, by applying graphic recognitiontechnology and risk identification

automation technology in risk assessmentprocess. It is one of specific application of using information technology (IT)to improve the work efficiency and accurate

of the traditional risk identificationprocess.

Designing andimplementing risk factors automatical acquisition. Firstly, the metro stationconstruction safety risk and risk factors are classified. Secondly, a relationtable of risk factors and engineering technical parameters which includesproject characteristics information, construction technical information geologyhydrological information, construction environment information, was built. Atlast, computer drawings recognition strategy and special algorithm weredesigned to capture various types of information recognition and impleteengineering semantics analysis, thus completing automatic extraction of theengineering technical parameters and risk

factors.

Building riskidentification knowledge base based on certainty factor. Theoretical knowledgeand expert experiential knowledge of metro station construction

safety risk identification was summarized, The metro construction safety risk and the risk factors are structured,the acquisition process for risk identification rules,

knowledge representationmethod and structured storage method of risk identification are interpreted, onthe basis of which the knowledge database based the certainty factors has beenbuilt.

Presenting twokinds of reasoning mechanism of casebased reasoning (CBR) and rulebasedreasoning (RBR) hybrid pattern. According to the characteristics  of the formationof risk identification rules, the rule based reasoning (RBR) based on the principleof fuzzy control is presented based on the element group retrieval reasoning algorithmand risk credibility calculation method. Summarize safety accident case of multilevelattributes, based on the casebased reasoning(CBR)and rule based reasoning (RBR)mixedmode in metro station construction safety accident database application model,and pointed out that in the premise of CBR leading RBR application method, realizesfrom the construction safety accident database acquisition risk identification information.

   Developing and applying the SRIS Theoverall framework, function models and database structure were designed andanalyzied. A case study of the Pangxiejia Station in Wuhan Metro Line 2 projectdemonstate the construction safety risk automatic identification process andrisk scenario validation.

   KeyWords: Metro station;  construction safety;  automatic risk identification;  construction drawings;  risk factor