工程质量风险分析与质量事故损失预测研究

王晓波

摘  要

改革开放以来，建筑业已经逐渐成为推动我国社会经济发展的支柱产业。但是在施工人员施工水平参差不齐、管理体制不完善、工程建设项目众多等因素的影响下，产生了大量的工程质量事故、引发了大量的工程质量问题。这些质量事故不一定会导致人员伤亡但是都可能给建筑企业造成损失，阻碍了社会经济的发展。工程质量事故描述文本中包含事故发生的部位、事故发生的原因以及造成的影响，但是以往对工程质量事故描述文本缺乏合理运用导致这些能够用来辅助决策的优质资源白白浪费。

所以本研究在收集工程质量事故案例的基础上，对工程质量事故进行统计分析和致因分析，然后在Bow-tie模型基础上进行工程质量风险分析，最后基于时间序列分析模型对工程质量事故损失进行预测研究。具体如下：

本研究以工程质量事故为研究对象，首先对工程质量事故案例进行统计分析，探寻出工程质量事故的发生特点以及发生规律，然后对工程质量事故案例进行致因分析，结合事故致因理论从人员、材料、机械和环境四个方面分析得出事故致因。基于LDA主题模型理论识别工程质量风险因素，建立工程质量事故故障树、事件树以及Bow-tie模型，最后提出工程质量风险规避措施为保证工程质量、减少甚至避免工程质量事故提供可供参考的依据。深入分析混凝土工程质量事故，建立事故原因故障树模型，求解最小割集和结构重要系数并且提出预防措施为保证质量、减少甚至避免发生混凝土工程质量事故提供依据。

本研究根据工程质量事故损失金额历史数据构建质量事故损失金额时间序列，随后检验序列的平稳性，由于数据序列平稳所以无需进行差分。然后对时间序列进行定阶并建立预测模型，最后检验预测模型并进行损失金额预测，为企业的决策提供依据，起到让企业预留储备金、尽早采取工程质量风险预防措施以及恢复措施以防工程质量事故导致损失的作用。

关键词：工程质量风险；工程质量事故；LDA模型；Bow-tie分析；时间序列分析

Abstract

Since the reform and opening up, the construction industry has gradually become a pillar industry to promote the social and economic development of our country. However, under the influence of the uneven level of construction personnel, imperfect management system, numerous construction projects and other factors, there are a large number of engineering quality accidents and a large number of engineering quality problems. These quality accidents may not necessarily lead to casualties, but they may cause losses to construction enterprises and hinder the development of social economy. The description text of engineering quality accident includes the location, cause and impact of the accident, but the lack of reasonable use of engineering quality accident description texts in the past leads to the waste of high-quality resources that can be used to assist decision-making.

Therefore, on the basis of collecting engineering quality accident cases, this study carries on statistical analysis and cause analysis of engineering quality accidents, then carries on engineering quality risk analysis based on Bow-tie model, and finally carries on the prediction research of engineering quality accident loss based on time series analysis model. The details are as follows:

This study takes engineering quality accidents as the research object, first carries on statistical analysis of engineering quality accident cases, explores occurrence characteristics and occurrence rules of engineering quality accidents, then carries on cause analysis of engineering quality accident cases. Combined with the accident cause theory, the causes of the accidents are analyzed from four aspects: personnel, material, machinery and environment. Based on the LDA theme model theory, the engineering quality risk factors are identified, and the engineering quality accident fault tree, event tree and Bow-tie model are established. Finally, the engineering quality risk avoidance measures are put forward to provide a reference basis for ensuring engineering quality, reducing or even avoiding engineering quality accidents. On the basis of in-depth analysis of concrete engineering quality accidents, the fault tree model of accident causes is established, and then the minimum cut sets and structural important coefficients are solved. Finally, preventive measures are put forward to provide a reference basis for ensuring concrete engineering quality, reducing or even avoiding concrete engineering quality accidents.

In this study, the time series of quality accident loss amount is constructed according to the historical data of engineering quality accident loss amount, and then the stationarity of the series is tested. Because the data series is stable, there is no need for difference. Then the order of the time series is determined and the prediction model is established. Finally, the prediction model is tested and the loss amount is predicted, which provides the basis for the decision-making of the enterprise. It plays the role of allowing enterprises to reserve reserves, taking engineering quality risk prevention measures and recovery measures as soon as possible to prevent losses caused by engineering quality accidents.

Keywords：engineering quality risk, engineering quality acciden, LDA model, Bow-tie analysis, time series analysis