Modularized fault diagnosis model of induction motor based on radial basis function neural network

Abstract

A modular fault diagnosis model of induction motor is proposed based on radial basis function neural network. The modular structure makes model configuration flexible, training time short and model convergence easier. The training algorithm of the fault diagnosis model is given. Through an example, factors of the affecting fault diagnosis model classification are analyzed, methods of feature extraction and feature enhance are discussed, and the algorithm of feature enhance is presented. Lastly, the construction, training and verification of the motor fault diagnosis model consisting of two sub-models are introduced. Research results show that because of adopting the modular model and using a sub-model to recognize a fault state, model training becomes easier. It is more important that the fault recognition ability and application flexibility of the fault diagnosis model proposed are improved obviously.