Fault Diagnosis of Vehicular Electric System Based on Neural Network

摘要

Analyzes the working features of silicon rectification generator, gives the reduced equivalent circuit and lists the mathematic model of the electric system. Based on the model, operates the typical fault in theory and normalizes the result. Then takes the normalized data as learning sample of the neural network. Compares the result of learning with either testing data or calculation data and both of them. Puts forward the fault mode of the electric system and neural network diagnosis mode, based on two methods. In one method, signals are analyzed directly by neural network. In another method, frequency analysis is performed before neural network analysis.