Research On The Mechanism Of How Design Failures Cause Unsafe Behavior-Based On Bayesian Network

Despite the increased safety awareness in the construction industry,the accident rate in the construction industry remains unacceptably high.Many studies focused on the reasons for construction accidents showed that human error played the most important role of accidents,thus,it is widely believed that the effective control of unsafe behavior can significantly reduce the possibility of accidents.With the proposal of safety design,both empirical and theoretical analyses indicated that design is the leading factor of unsafe behavior.However,current studies usually focused on the correlation between unsafe behavior and design features,and contributed little to how design affects human behavior.In addition,the existing models were not able to specify the mechanism of the changed environment influence on the cognitive factors and then effect on human behavior.As a result,the stakeholders are not able to understand the importance of safety design,and they will hold a negative attitude towards the application of safety design.To overcome the questions mentioned above,we would build a human error causal model of construction industry in order to specify the mechanism of how design affect unsafe behavior,and quantify the design's impact on unsafe behaviors based on real data,then identify the most possible path between design and unsafe behavior,and finally interpret the detailed impact process.Four detailed outcomes was produced in this research: 1)Based on antecedentconsequent relationship in CREAM method and interview with experts panel,the relationship and definition of variables were modified according to the characteristics of construction industry,and theoretical human error causal model was built.2)We used the real safety inspection data as data source,and conducted the Bayesian structural learning in order to build the real-site human error causal model of elevator installation.3)We used the previous model as the network and conducted the Bayesian parameter learning to calculate the conditional probability of unsafe behavior and other nodes in the network.4)We identified all the possible influence paths from starting nodes to unsafe behavior,and quantified the impact of every path.According to the cross analysis between design and other three staring nodes,we can acquire the critical path between design and unsafe behavior.The finally result indicates that when design failure(K4)occurred,the occurrence likelihood of unsafe behavior increased by 80%.When design failure occurred,it will create an unfamiliar circumstance,and affect the diagnose process of cognitive activity,finally bring a change on the human behavior.In general,the human error causal model and quantitative result are born from real data,and the mechanism of how design affect cognitive factors and then human behavior was specified.Besides,a possible negative path was found as well.This study proves the importance of safety design and interpret the mechanism between design and behavior,which contributes to the promotion of safety design.