Research On Induction Motor Fault Monitoring Based On Genetic Algorithm-Neural Network

This paper discusses the characteristics of induction motor fault diagnosis and its achievements in recent academic research, which is based on a large collection of the abroad and domestic scientific and technical documents. The diagnostic system is designed with the help of genetic algorithm-neural network (GA-NN) to monitor and diagnose whether an induction motor performs normally by detecting the stator current.The stator-winding short circuits fault, broken rotor bar fault, eccentricity fault and bearing fault are firstly analyzed in detail in the case of induction motor based on its operational principle. The typical frequencies and their relevant amplitude are extracted and the inherent relations between motor faults and typical frequencies are presented. The typical signals of various motor faults are simulated effectively with software and hardware.To overcome the drawback of BP algorithm and single neural network diagnosis model, a GA-NN algorithm is proposed, which can evidently accelerates the convergent speed, reinforce the generalization ability than usual and improve the fault diagnosis effect.According to different kinds of motor faults, the fault samples are collected to train the GA-NN. The neural network after training acts as the final diagnostic device. A great deal of simulation work has been carried out with this intelligent diagnosis device and better results are received which verify a effective fault diagnosis method with higher veracity and better generalization abilityThe experiment researches are carried out with advanced networks in ROCKWELL laboratory. A flexible motor faults simulation system is composed using a computer with a hi-speed data gathering card and the data exchange is implemented with Socket functions of VC++. The intelligent diagnosis scheme proposed is validated and real-time fault diagnosis via networks is realized.