Rainfall class prediction model based on PSO-BP neural network

In recent years, Guangzhou City has frequently suffered from grid disasters caused by typhoons and rainfall, and the operation of the power system under extreme weather is facing severe challenges, and the accurate prediction of rainfall level has become a key topic to reduce disaster losses. The traditional BP neural network is difficult to achieve the global optimal solution, which restricts the generalization ability of the model. To this end, this paper proposes an improved BP neural network model based on particle swarm optimization (PSO) algorithm, based on the daily-valued meteorological data of Guangzhou City from 2020 to 2024, and selects the precipitation factors such as humidity, temperature, wind speed, etc., and optimizes the BP network weights and thresholds using the PSO global search mechanism to construct the rainfall level prediction model. The experimental results show that compared with the traditional BP algorithm, the PSO-BP model has an accuracy of 85.7% in precipitation level prediction, which significantly improves the prediction performance. This study provides a high-precision meteorological prediction scheme for power system disaster prevention under extreme weather, which is of practical significance for urban disaster risk management.