Early Warning of Construction Workers’ Safety Behavior from Cross-level Perspective

Abstract:

Based on the theory of behavior safety, the literature analysis and personnel interviews, the early warning system for the safety behavior of construction workers was established from a crosslevel systematic perspective. According to the system, the questionnaire was designed, the data were collected from construction sites, and the system was trained and tested by back propagation (BP) neural network. The results show that output values of the simulation accord well with the actual safety behavior data, and the simulation ability of the trained neural network model is competitive. The evaluation system could effectively evaluate and predict the safety behavior of the construction workers. At the same time, safety behavior was sorted into three grades (excellent, good and poor) according to the threshold, and different safety intervention programs should be adopted to enhance the safety management ability and improve the projects’ safety performance.