AI-Enabled Rule Extraction System for BIM-based Fire Performance Diagnostic Platform (AI-FireBIM)

对不幸的格伦费尔火灾的审查揭示了建筑监管制度的尖锐做法，这在21世纪世纪的英国被描述为令人震惊的。应对挑战，确保每个人都做正确的事情，并追究那些试图偷工减料的人的责任。根据这一点，朱迪思·哈克特女士领导的团队建议需要一个新的智能系统，用于高层和复杂建筑的监管和执法。_该系统预计将促进高度自动化，以评估和诊断是否符合一系列消防安全法规，如_批准文件B、HTM 05-02、BS 9999、BS 9991和BB 100，其中一些法规自Grenfell火灾调查后已通过审查程序。为了支持向自动化消防安全合规性诊断迈进，我们的联合体最近完成了Fire-BIM项目，该项目评估建筑设计，确定不符合要求的区域，并提出设计变更建议，以满足消防安全要求。Innovation Fire Engineering、P+HS Architects、Anglestack Limited、利兹贝克特大学、Farrans Construction PLC、CLM咨询公司和赫特福德郡大学之间的合作导致了FireBIM平台的开发，该平台是消防安全合规性诊断的自动化门户，目前正由一家分拆公司开发。然而，我们的FireBIM系统是基于规则的设计，涉及手动提取消防安全规则，并将其编码到MongoDB中作为NoSQL数据库。因此，本可行性研究为使用自然语言处理（NLP）作为人工智能技术自动捕获建筑物安全规则提供了机会。可行性研究将用于确定执行自动规则捕获的最佳方法及其作为决策支持系统的实现，包括文本预处理，句子分割，命名实体识别（与建筑构件和元件一致），通过模式匹配或基于规则的解析进行规则识别，规则提取和实体捕获，规则分类和归类，通过与现有手动提取的规则数据库进行比较，确保规则质量。可行性研究的结果将支持我们的FireBIM诊断平台自动生成规则，从而最终形成AI-FireBIM工具。这将通过以下方式评估建筑物设计是否符合批准文件B：（i）自动诊断建议/竣工设计是否符合目标法规/标准的规定。㈡查明不符合条例和标准的领域。㈢提出所需的设计修改和不同组成部分的消防安全标准要求。

The review into the unfortunate Grenfell fire disaster uncovered sharp practices in building regulation regime, which is described as shocking in the 21st century United Kingdom. To tackle the challenges and ensure that everyone does the right thing and those who try to cut corner are held accountable. In line with this, Dame Judith Hackett led team suggested the needs for a new intelligent system of regulation and enforcement for high-rise and complex buildings. _Such system is expected to facilitate a high level of automation to evaluate and diagnose compliance with the array of fire safety legislation such as the_ Approved Document B, HTM 05-02, BS 9999, BS 9991 and BB100, some of which has gone through review processes since the aftermath of Grenfell fire enquiry.To support the move towards automated fire safety compliance diagnostic, our consortium has recently completed Fire-BIM project, which evaluates building design, identify areas of non-conformances, and suggest design changes for meeting fire safety requirements. The collaboration among Innovation Fire Engineering, P+HS Architects, Anglestack Limited, Leeds Beckett University, Farrans Construction PLC, CLM consultancy and University of Hertfordshire has led to the development of FireBIM platform as an automated portal for fire safety compliance diagnostic, which is currently being exploited by a spin out company. However, our FireBIM system is a rule-based design, involving manual extraction of fire safety rules that is coded into MongoDB as a NoSQL database.Thus, this feasibility study provides opportunity for automatically capturing building safety rule using Natural Language Processing (NLP) as a technique in Artificial Intelligence. The feasibility study will be used to determine the optimal approach for executing automated rule capturing and its implementation as a decision support system, covering text pre-processing, sentence segmentation, named entity recognition (in line with building components and elements), rule identification through pattern matching or rule-based parsing, rule extraction and entity capturing, rule classification and categorization, and rule quality assurance through comparison with existing manually extracted rule database.The result of the feasibility study would support in automating rule generation for our FireBIM diagnostics platform, thereby culminating in AI-FireBIM tool. This would evaluate building design for compliance with Approved Document B by(i)Automatically diagnosing proposed/as-built designs for compliance with provisions of targeted regulations/standards.(ii)Identifying areas of non-conformances with regulations and standards.(iii)Suggesting required design changes and fire safety standard requirements for different elements.