Research On Decision Rules Of Hazardous Chemical Accidents Cases Oriented To Accident Analysis

In recent years,the Chinese government has adopted a series of major hazardous chemical safety measures,and the number of accidents has shown a steady decline.However,hazardous chemical accidents still occur frequently in China.Emphasis on accident cases with a wealth of knowledge of the event evolution and in-depth analysis of accidents that have occurred,some unknown but necessary knowledge can be obtained to find the root cause of its occurrence,and then fundamentally change the current situation of frequent accidents.This knowledge can not only be used to assist decision makers to make rapid and scientific decisions,but also help to understand the mechanism of hazardous chemical accidents,so as to achieve the goal of accident prevention.Compared with foreign studies,the current analysis of hazardous chemical accidents in China is mainly based on the frequency analysis of a single descriptive factor,and lacks the discussion of the interaction of multiple factors.Although statistical analysis can be used to infer simple causal relationships between these factors,the large number of factors involved makes it difficult for researchers to identify some important associations between related factors in hazardous chemical accident data.In addition,the existing research on rule mining regards its occurrence as a single-stage of "cause-effect" model,and lacks detailed exploration of the evolution of accidents.Rule mining based on different accident stages will further analyze hazardous chemical accidents.In response to the above problems,this paper proposes a multi-layer weighted decision rule mining model,combined with complex network theory,to explore the potential causality and the occurrence rules hidden in hazardous chemical accident data.First of all,this paper takes 239 hazardous chemical accident cases as a qualitative analysis resource of the grounded theory,and proposes an accident causation framework based on the accident life cycle referring to the existing accident causuation models and the unique process safety management theory of the hazardous chemical industry.Then,a multi-layer weighted decision rule mining model is constructed,and the Apriori algorithm is optimized from the two perspectives of single stage and cross-stage to satisfy the requirement of the mining forms of "accident causation set(conditional attributes)?accident type(decision attribute)" and "accident causation set(incubation period)? accident causation set(occurrence/expansion period)?accident type(decision attribute)".Finally,this paper use complex network theory to study the system structure characteristics and nonlinear relationships between accident types and accident causing factors in the linear decision rules for hazardous chemical accidents,Combining its linear and non-linear correlations and summarize the rules of various hazardous chemical accidents and propose targeted prevention measures combinng their linear and non-linear correlations.The results of these analysis can provide management and employees with evidence of possible causality,thereby helping to establish a safer work environment.