基于控制语句自动获取的建筑工程质量表象与对应原因分析  

Analysis of Construction Quality Appearances and Corresponding Causes Based on Automatic Acquisition of Control Statements

作　　者：张明媛 贺凯 刘运鹄 周光毅[2] ZHANG Mingyuan;HE Kai;LIU Yunhu;ZHOU Guangyi(Faculty of Infrastructure Engineering,Dalian University of Technology,Dalian,Liaoning116024,China;Northeast Branch of China Construction Eighth Engineering Division Co.,Ltd.,Dalian,Liaoning116021,China)

机构地区：[1]大连理工大学建设工程学部,辽宁大连116024 [2]中国建筑第八工程局有限公司东北分公司,辽宁大连116021

出　　处：《施工技术（中英文）》2022年第20期39-44,共6页Construction Technology

基　　金：辽宁省自然科学基金(优秀人才培育项目):基于机器视觉和群智感知的施工现场危险场景识别与预警研究(2019-MS-052)。

摘　　要：质量管理工作在施工前的技术方案设计阶段即已开始,在工程实施过程中及实施后产生的质量问题,也常以质量报告文本形式交付专家进行阅读并分析判断,根据经验给出引起质量问题的可能原因,而这一过程可通过人工智能技术进行自动化实现,提高管理效率。为实现这一自动化过程,建立质量问题表象与质量保障规范的对应关系,是首要解决的问题。基于建筑工程施工质量验收规范,构建适用于工程质量保障的规范文本集;借助Python数据分析工具,以分部分项工程为分类依据,将质量规范文本进行自动化分类,获得保障工程质量的控制语句;以63份实际工程质量问题报告为样本,梳理出15类质量问题表象;通过专家访谈,建立控制语句与问题表象之间的对应关系;以地下室渗漏质量问题为例,验证上述方法的可行性。该方法能有针对性地自动快速锁定产生质量表象问题对应的质量规范控制语句,获得不满足质量规范的施工技术措施条目,进而分析出可能产生的质量问题原因。这一过程可弥补自动化实现方法在实际工程质量检查环节的应用空白,为提高工程项目质量管理效率提供方法依据。Quality management work starts at the design stage of technical solutions before construction, and quality problems arising during and after the implementation of the project are often delivered to experts in the form of quality report texts for reading, analyzing and judging. Then experts give possible causes of quality problems based on experience. This process can be automated by artificial intelligence technology to improve management efficiency. In order to realize this automated process, establishing the correspondence between the appearance of quality problems and quality assurance specifications is the primary problem to be solved. Based on the construction quality acceptance specification of construction projects, specification texts set applicable to project quality assurance were constructed. With the help of Python data analysis tool, quality specification texts were automatically classified based on the classification of subdivision projects to obtain control statements for guaranteeing project quality. 63 actual project quality problem reports were used as samples to sort out 15 types of quality problem appearances. Through expert interviews, the correspondence between control statements and problem appearances was established. The feasibility of the above method was verified by taking basement leakage quality problem as an example. The method can automatically and quickly target the quality specification control statements corresponding to the quality appearance problems, obtain construction technical measures that do not satisfy the quality specification, and then analyze the possible causes of quality problems. This process can fill the application gap of the automated implementation method in the actual engineering quality inspection process and provide a methodological basis for improving the efficiency of engineering project quality management.

关 键 词：工程质量 自动化 控制语句 问题表象 地下室 渗漏 

分 类 号：TU712.3[建筑科学—建筑技术科学]