Application Research Of Accident-Causing Analysis And Prediction Based On Building Collapse Case Base

With the rapid development of our country’s economy,the Construction Industry,as one of the pillar industries of the national economy,is also booming.The construction industry is characterized by strong fluidity,low mechanization,high risk coefficient and so on.It is an industry that tends to cause safety accidents.In particular,in recent years,various new technologies and processes have emerged one after another,the construction difficulties have been increasing continuously,and safety production accidents have occurred from time to time.Therefore,it is of great significance to analyze the causes of building collapse accidents and establish a long-term accident prevention mechanism for architectural engineering to reduce the occurrence of building collapse accidents.Based on the perspective of Building Collapse Safety Management,this thesis analyzes 251 investigation reports of Building Collapse accidents,and studies the causes of Collapse Accidents based on Accident Grades and Types.Through the study and analysis of the accident report,the historical literature review and the interview and investigation of experts and scholars,using the 24 Model,the four levels of causes,14 contributing factors and 45 sub contributing factors leading to the Building Collapse Accident are summarized,thus constructed the Building Collapse Cause Identification Index System.On this basis,based on the eight types of Collapse Accidents,the Grey Relational Analysis Method is used to seek the relation between the Accident Causes and The Accident Grades and Types,sort the accident causes according to the Correlation Degree,and make targeted preventive measures according to the Correlation Degree.Then,the Bayesian Network Model is constructed based on the Factor Correlation Degree,and then the key factors,sensitive factors and Most Probable Paths that lead to Building Collapse Safety Accidents are determined through Diagnostic Reasoning and Sensitivity Analysis.On the basis of the initial data set,the accuracy of the Bayesian network model is verified by Python 3.0.In addition,taking each cause factor of Building Collapse Safety Accident as sample data,and carrying out Preliminary Data processing,the prediction model of Building Collapse Safety Accident based on Random Forest is constructed.Three parameters,namely,the number of Decision trees,the depth of Decision trees,and the minimum number of samples that can be divided into nodes,are selected to debug the model,and the accuracy of prediction of Building Collapse Safety Accident Grades and Types is empirically analyzed by using Random Forest algorithm.This thesis takes the Building Collapse Accident Case Database as The Research Foundation,gives full play to the application value of the accident investigation report,fully extracts the effective information of the Accident Investigation Report,analyzes and obtains the Basic Laws Of The Building Collapse Accident and the relationship between the cause and the accident,and establishes The Accident Prediction Model according to the research results,and puts forward the Management Countermeasures,which provides a reference for the Prevention Of The Building Collapse Accident.This thesis consists of 41 figures,28 tables and 83 references.