Characterization And Prediction Of Unsafe Behaviors Of Construction Workers Based On Cognitive Science

In the past five years,the number of safety accidents and deaths in the construction industry has increased year by year,which shows that it is exceptionally urgent to solve construction safety.Unsafe behavior is the leading cause of safety accidents.However,the existing unsafe behavior management measures have application limitations,which can not solve the unsafe behavior problems in the construction process.In addition,the existing management measures treat construction workers as a whole while ignoring the problem that the differences of individual characteristics will form different groups and influence the management measures of unsafe behaviors.Therefore,this paper aims to build a security refinement management mode focusing on early warning of unsafe behaviors and oriented by individual characteristics.The main research work and achievements of this thesis are as follows:(1)Based on the cognitive science research paradigm and the cognitive model proposed by predecessors,11 fundamental causes of cognitive failure are obtained through literature review and induction to construct a more systematic and comprehensive cognitive model.(2)Eleven factors based on the cognitive model are used as the theoretical basis for developing the questionnaire.Meanwhile,the questionnaire of unsafe behaviors of construction workers based on the cognitive model with high reliability and validity is obtained using the methods of deviation avoidance,questionnaire optimization,and questionnaire pre-investigation.(3)Because there are significant differences in individual cognition,11 factors based on the cognitive model are used as individual characteristic indicators of unsafe behaviors of construction workers.Through potential category analysis,the unsafe behaviors of construction workers and their potential characteristics are determined.The results show that the unsafe behaviors of construction workers can be divided into five categories:cognitive excellence,knowingly committing crimes,cognitive abnormality,overconfidence,and lack of knowledge.(4)Mark the categories of data one by one according to the analysis results of potential categories.Six machine learning algorithms and traditional modeling techniques are used to train,verify and test the labeled data.The results show that the accuracy,accuracy,and recall rate of support vector machines are the highest.(5)The early warning mechanism of unsafe behavior of construction workers is selected by support vector organization,and an empirical study is made on the early warning mechanism of unsafe behavior.The unsafe behaviors of four construction workers were observed by behavioral observation and verified by independent questionnaires,multiple interviews,and category comparison.The results show that the early warning mechanism constructed in this paper can be used to measure the unsafe behaviors of construction workers.(6)According to the potential characteristics of unsafe behaviors of five groups of construction workers,the corresponding safety management measures are put forward based on the theoretical basis of management and behavioral science.On the theoretical level,based on cognitive science,this paper reveals the cognitive mechanism of unsafe behavior of construction workers and constructs a more systematic and comprehensive cognitive model,which provides a new perspective for the study of unsafe behavior.Furthermore,the unsafe behavior management tools are developed for construction enterprises on the practical level,which has significant application value for improving construction workers’ unsafe behavior management.