Application of Bayesian optimization-based cloud model in fire risk assessment of distributed photovoltaic power plants
Yukun Zhao, Tingting Luan, Xiaoyun Li, Kai You Wang, Shang Shi · Energy Sources Part A Recovery Utilization and Environmental Effects · 2025
With the large-scale deployment of distributed photovoltaic power stations, their fire hazards are becoming increasingly prominent. Aiming at the shortcomings of existing risk assessment methods in dealing with complexity and uncertainty, this paper proposes an improved cloud model based on particle swarm optimization and Bayesian optimization to more accurately assess the fire risk of distributed photovoltaic power stations. First of all, with the help of fishbone diagram to identify fire incentives, build a scientific risk index system; secondly, the subjective and objective weights are calculated by combining the analytic hierarchy process and the entropy weight method, and the adaptive fusion of weights is realized by particle swarm optimization to improve the rationality of weight distribution. Subsequently, the cloud model is used to quantify the uncertainty and ambiguity of the risk, and Bayesian optimization is introduced to globally optimize the cloud model parameters to improve the stability and accuracy of the model. Finally, based on the optimized model, an actual power station is evaluated. The results show that the overall risk is “General Risk,” among which “Regional Environment” has the highest risk value, accounting for 0.8649, and is the key risk source. After the implementation of targeted safety measures, the risk level is reduced to “Low Risk,” which verifies the scientificity and practical value of the proposed method.