Fault Diagnosis And State Evaluation Of Induction Motors Based On Adaptive Filtering Of Electrical Signals

Modern Manufacturing and Operation is reducing the reliance on human being,in the meantime,more and more depend on the automation equipment.As a widely used industrial facility,motor is an integral part of the modernization mass production.Fault diagnosis system of motor is becoming significant,playing more and more important role in safe operation.The goal of fault diagnosis system is to ensure safety in the motor operating.During the development process,signal processing is a critical component,most researches are centered in this part.This paper introduces the main fault types of the motor,compares the current mainstream fault diagnosis techniques,and describes and discusses the stator current method.In the important signal processing part of the diagnostic system,this paper adopts an improved adaptive filter based on correlation algorithm.In MATLAB,the stator current signal analysis method based on the improved adaptive filtering method is used for diagnosis and verification.Successfully separated the characteristic frequency of the fault category and detected the fault type.The feasibility of the improved induction motor fault diagnosis scheme proposed in this paper is verified.In details,this paper introduces the research background and the significance of motor fault diagnosis.The development of the domestic and foreign motor fault diagnosis techniques is outlined.Specially,we analyzed data and relevant statistics about the distribution of fault types and the reasons of the failure.Second,the mathematical principle of the motor model is elaborated.A mathematical model for parameter identification of asynchronous machines was established.The SA—PSO method was used to identify the parameters of the normal motor,and parameters of common motor faults were identified.From the experimental results,we can see the changes of the motor parameters under different faults.Further,a method for evaluating the performance degradation of a motor under failure data is studied.The motor characteristic signal is input to the SOM neural network fortraining,an evaluation model of the SOM network is established,and the MQE value is calculated by the evaluation model to evaluate the degradation state of the equipment performance.Finally,the fault test of the motor bearing is used to verify the validity of the method.It can describe the various stages of motor fault performance degradation.Finally,the GM—LSSVM combination forecasting model is studied,that is,a combination forecasting method based on GM model and LSSVM model is proposed.Finally,based on the actual motor health information data,the corresponding predictive evaluation criteria were obtained through simulation analysis.