Cause ZIP model of the lifting injury accident based on the coupling influence of mean-and-correlation
In recent years,lifting injury accidents have continued to be common,resulting in significant direct economic losses and casualties.These accidents have received high attention from the industry.To explore the level of influence of the causes of lifting injury accidents,this study proposes a hierarchical management and control of accident risk for the safety management of lifting operations after collecting the accident reports and establishing a Zero-Inflated Poisson(ZIP)regression model.A 24Model and system safety analysis method are used to identify accident causes from 376 investigation reports of lifting injury accidents from China(2011-2021).27 causes are identified to form the accident cause system.The Poisson model and the ZIP regression model of the causes of lifting injury accidents are constructed with the accident classification as the dependent variable and 27 accident causes as independent variables.The Vuong method and Goodness of fit test criteria are used to test the two models.By comprehensively considering the average frequency of accident causes and the correlation between causes and accidents,the elastic analysis method is used to quantitatively rank the degree of impact of accident causes.The research results show that the fit of the ZIP regression model is better than that of the Poisson regression model.The degree of influence of each cause on the lifting injury accidents is ranked as follows:inadequate safety supervision C6(Tk=1.0221),weak safety awareness H1(Tk=0.5117),illegal operation of operators H3(Tk=0.4758),unlicensed employment of operators H2(Tk=0.2116),unqualified special construction plan C7(Tk=0.1234),damaged structural components or unreliable connection of structural components M3(Tk=0.1045).This study characterizes the impact of accident causes on lifting injury accidents from two dimensions-the average frequency of accident causes and the correlation between causes and accidents,providing new research ideas and methods for risk classification and control of lifting injury accidents.
万方数据
维普
