Volume 48

Fault Diagnosis Method for Charging Pile Based on Improved BP Neural Network Yang Fang, Yang Fengkun, Chen Liangliang, Zhao Yue, Xu Ke

Abstract

In order to ensure the safe and stable operation of electric vehicle (EV) charging stations and improve the accuracy of fault diagnosis, a fault diagnosis method based on an improved Backpropagation Neural Network (BP) is proposed. This method first preprocesses the operational dataset of the charging stations. Then, the preprocessed dataset is input into the BP model for training to learn the correlation between the normal and faulty states of the charging stations. Finally, an improved optimization technique is introduced to optimize the weights and thresholds of the BP model. This technique combines the Firefly Algorithm (FA) and the Northern Goshawk Optimization Algorithm (NGO) to obtain the optimal model by optimizing the BP model. Simulation results demonstrate that the proposed improved BP method has good computational advantages in terms of precision and recall rates. Compared to the traditional BP algorithm, the improved BP method achieves a 10.83% increase in diagnostic accuracy and can accurately diagnose the status of the charging stations

Keywords charging pile, fault diagnosis, neural network, firefly algorithm, northern goshawk optimization