Life Prediction of Capacitor Based on AVC by ESM-BP Hybrid Neural Network Model

Abstract

In order to better prevent power capacitor trip breakdown in power system and improving maintenance efficiency of power capacitor, a hybrid model based on ESM (Expert Scoring Method, ESM) and BP (Back Propagation, BP) neural network is proposed for capacitor life prediction after deeply analyzing mass effective data of power capacitors, which aiming at the characteristics of capacitor trip under AVC (Automatic Voltage Control, AVC) control strategy. The input of the ESM-BP hybrid neural network model, using to training the model to predicting the capacitor life, is the trip data of 177 power capacitor banks in the east area of City D in Guangdong Power Grid. The testing data of the model is the trip data of 10 power capacitor banks in the east area of City D. The test result shows that the ESM-BP hybrid neural network model owns high prediction accuracy. The prediction method proposed in this paper can be widely used to prediction the lifetime of power capacitors.