Research On Safety Risk Transmission In Urban Rail Transit Construction Through Data Mining

Urban rail transit construction projects have the characteristics of large scale,long cycle,multiple participants,complex underground environment,and considerable social attention.Therefore,the construction is difficult and with high safety risks and safety accidents happened.On the one hand,the accident investigation reports contained a large amount of information on safety risks,but they were shelved and not used effectively.On the other hand,the traditional safety risk management selects independent risk factors as the research object,relies on the subjective experience of experts to establish risk analysis models,evaluates the risk factors and formulates risk response measures or emergency plans.This process neglects the transmission behavior among risk factors and is subjective.Thus,it is difficult to adapt to the management needs of the safety risks of urban rail transit construction.Based on this,this study discusses the safety risk transmission relationship and evolution processes in urban rail transit construction through data mining in order to promote the continuous improvement of the safety risk management of urban rail transit construction projects.Specific research content includes:The features and structure of the safety risk system of urban rail transit construction projects are analyzed on the basis of system engineering theory.The source,concept,type,and mathematical expression of safety risk transmission are proposed and the safety risk transmission of urban rail transit construction projects is formulated.This work lays the theoretical and methodological foundation for analyzing the safety risk transmission in urban rail transit construction projects.With the safety accident investigation reports of urban rail transit construction projects as the corpus,text mining is conducted to extract the safety risk factors that used to cause safety accidents.A suitable professional lexicon is developed to achieve more accurate text segmentation.Information entropy is introduced to propose a feature selection method on the basis of the cumulative entropy weight term frequency to screen high-frequency words that characterize safety risk factors.Then,safety risk factors of urban rail transit construction projects are listed and classified.This process realizes the structural transformation of the accident investigation reports from the text.In comparison with the national standard norms,the rationality and comprehensiveness of the safety risk factors are verified.The association rules and natural language process(NLP)technology is adopted to extract the casual relationships and coupling relationships among safety risk factors.Under the framework of the association rules of Support and Confidence,an interesting rule evaluation model with multiple indicators is constructed to snap the associated safety risk factors from big data.Then,the causal and coupling relationship extraction patterns are proposed on the basis of semantic dependence with the NLP technology.The patters help the computer to understand the relationships hidden in the accident investigation reports.Then a numerical transmission matrix is formed.Finally,the method of the Decision-Making Trial and Evaluation Laboratory(DEMATEL)is used to analyze the risk transmission strength in four dimensions,namely,influence degree,affected degree,centrality degree,and causality degree,to reveal the transmission strength among the safety risk factors of urban rail transit construction projects.A safety risk assessment model is formulated on the basis of the Bayesian network(BN)structure to research the status of urban rail transit construction projects under the effect of safety risk transmission.The interpretive structural model(ISM)is formulated and is transformed to Bayesian topological structure of the safety risk network.The probability parameters of Bayesian network are automatically generated by machine learning,which makes up for the shortage that prior probability in Bayesian network mainly depends on expert judgment.Then,causal reasoning is used to analyze the safety risk status of urban rail transit construction projects.Diagnostic reasoning is applied to evaluate the probabilities of safety risk factor occurrence and extract the critical risk transmission paths.The assessment model reveals the evolution process of the safety risk system of urban rail transit construction projects under the effect of safety risk transmission.Finally,the three dimensions of risk transmission strength,safety risk occurrence probability,and safety risk system structure level are combined.The rank criteria of safety risk assessment based on the importance–urgency matrix are formulated.The grading and precise safety risk response strategies are put forward separately on the basis of the anti-transmission conception,which provides a new approach to safety risk response from the perspective of anti-transmission.Finally,the empirical and applied analyses are conducted in the X project.The research shows that the proposed risk transmission-based safety risk assessment model can better simulate the evolution process among multiple safety risk factors and verifies that the applicability of data mining method in the safety risk analysis of urban rail transit construction projects,which can provide effective safety risk assessment and prediction for all units in urban rail transit construction to improve the efficiency of safety risk response.The paper has a total of 104 pictures,65 tables,and 206 references.