Research On Unsafe Behavior Early Warning Of Construction Workers Based On BP Neural Network

The construction industry is a prominent industry that promotes the high-quality development of China’s economy and it is closely related to people’s production and life.However,in recent years,construction safety accidents occur from time to time,resulting in huge casualties and economic losses,China’s construction safety production situation is still grim.Relevant studies have shown that unsafe behavior is one of the main causes of accidents.Therefore,starting with the prevention and control of unsafe behaviors,identification and early warning of unsafe behaviors,so as to take prevention and control countermeasures in advance and reduce the occurrence of safety accidents are of great significance to individuals,enterprises and even the society.In this paper,construction workers as the research object,through the study of domestic and foreign related literature,analyzing the research status of behavioral safety management theory,influencing factors of unsafe behavior,prediction and early warning of unsafe behavior,and prevention and control countermeasures of unsafe behavior.Under the premise of complex influencing factors,insufficient prediction and early warning validity and unclear specificity of prevention and control countermeasures,five methods of accident statistical analysis,case analysis,literature analysis,qualitative interview and questionnaire survey were adopted to obtain the evaluation index of unsafe behavior.The evaluation index system of unsafe behavior composed of 4 first-level indexes and 23 second-level indexes is established by consulting experts.On this basis,based on the principle of BP neural network,23 evaluation indicators are taken as network input,and four unsafe behavior states,namely,safety,light alarm,medium alarm and heavy alarm,are taken as network output.Through the repeated training and learning of the questionnaire data,the "23-9-4" three-level structure of unsafe behavior warning model was finally constructed,and the model was trained,verified and tested.The results show that the model has good warning accuracy and strong fit with the data,and the weight of the evaluation index of unsafe behavior can be obtained according to the trained model.According to the weight value,specific prevention and control countermeasures are put forward from the four aspects of individual,organization management,equipment and external environment,so as to provide reference basis for enterprises to prevent and control unsafe behavior.