In recent years,with the rapid development of China’s construction industry,frequent construction safety accidents have caused numerous casualties and economic losses,affecting social harmony,stability,and economic development.The falling from height(FFH)accident has a high frequency and leads to numerous deaths,which is known as the first of the "five major injuries" in the construction industry.Therefore,it is of great significance to analyze the correlation between the accident causes of FFH,find out the critical causal chain that is most likely to cause the accidents,and formulate targeted preventive measures and safety management strategies according to the critical causal chain,for reducing the risk of FFH and enhancing the level of construction safety.In this thesis,the Bayesian Network(BN)of FFH accidents is constructed by combining knowledge and data.Knowledge provides structural restrictions for model construction and enhances the interpretability of the model.Data provides the objective basis for model construction and ensures the accuracy of the model.In terms of knowledge acquisition,the accident analysis framework of FFH is built based on the STAMP model.The framework can explain the safety control mechanism of the system from macro and micro levels,and explain the accident causes and the correlations among them according to the thought that safety control failure leads to accidents.In terms of data acquisition,an accident cause classification framework is designed with the advantages of natural language processing technology,which improves the efficiency of obtaining accident cause categories from accident investigation reports,and then the structured accident cause data is obtained.Finally extracting the structural restrictions from the accident analysis framework,taking structured accident causes as data,the BN model of FFH accidents is built based on the structural restrictions.The BN describes the causal relationship among the accident causes qualitatively and quantitatively and deduces the critical cause chain that has a significant impact on the accidents.Based on the research method in this thesis,26 cause categories of FFH are determined,the qualitative and quantitative causal relationships among accident causes are analyzed,and the critical causal chain of FFH is mined.Finally,according to the critical causal chain,accident prevention suggestions are put forward to provide references for the safety management of relevant subjects,so as to reduce the occurrence probability of FFH accidents,ensure the safety of the construction process,and avoid casualties and property losses. |