Causal Analysis of Accidents During Wind Power Engineering Construction Using Bayesian Networks

Abstract

To investigate the causes of accidents during the construction phase of wind power projects and to prevent their occurrence, this study draws on accident investigation reports and the "2-4"Model to summarize unsafe behaviors and actions. It selects causative variables from four aspects: human, equipment, environment, and management. Integrating fault tree analysis, a Bayesian network (BN) model for analyzing the causes of accidents during wind power engineering construction is constructed using the BN software, GeNie. The model undergoes structural and parameter learning, calculating the conditional probability distribution and posterior probabilities of each node. Through variable sensitivity and analysis of the most significant causative chains of accidents, the key factor paths leading to accidents are identified, contributing to reducing the accident rate during the construction phase of wind power projects. The results indicate that inadequate personal protection and violations of regulations are prevalent among human factors. In terms of management factors, insufficient safety management and supervision are the main contributors to accidents, with a probability value exceeding 70%. Geological conditions, road conditions, limited workspace, exceptional environmental changes, proximity to energized machines, and safety protection equipment failures are significant factors in accidents during the construction phase.