Research On Fault Diagnosis Of Asynchronous Motor Based On Three-phase Stator Current Multi-characteristics

Asynchronous motors are the most widely used electromechanical energy conversion devices and terminal energy-consuming equipment in industrial and mining fields.However,the asynchronous motors are prone to various failures by harsh working environment and longterm overload wear and other problems.Therefore,conducting intelligent and reliable motor fault diagnosis research have crucial theoretical purport and economic value for guaranteeing the quality of production and operation,ameliorating the working efficiency of equipment and reducing energy consumption.Aiming at the problem of multi-fault diagnosis of asynchronous motors,this paper focuses on the fault diagnosis of broken rotor bars,air gap eccentricity and mixed faults(broken rotor bars and air gap static eccentricity faults)in asynchronous motors.Firstly,the fault principle of the motor is analyzed and the characteristic frequency which appears in the stator current is deduced.The motor models under normal and fault conditions were established by ANSYS Maxwell finite element simulation,the corresponding three-phase stator current characteristic signals were obtained,and the correctness of the extracted stator current signal data was verified by analyzing the current spectrum of the motor under different fault conditions.At the same time,it provides reliable data support for subsequent fault feature extraction.Secondly,due to the problem of fundamental frequency spectrum leakage,it is difficult to diagnose rotor faults by classical spectrum analysis.The research method based on the combination of the stator current Park vector distortion rate feature and the wavelet packet transform to extract the vector mode energy feature is proposed in this paper.By analyzing the theoretical method of Park vector transformation in motor fault diagnosis,it is verified that the characteristics of inner and outer trajectory distortion rate obtained by Park vector transformation can clearly reflect the ch aracteristic changes under different loads.On this basis,the extended Park vector transform is used to convert the three-phase stator current signal into a one-dimensional time series signal,and the wavelet packet coefficient energy feature extraction is carried out after eliminating the influence of the fundamental frequency component in the signal.The selection of wavelet basis functions and the uncertainty of the number of decomposition layers are analyzed and studied from the perspectives of similarity and fault spectrum characteristics.Finally,by analyzing the advantages and disadvantages of Park vector distortion rate feature diagnosis results and wavelet packet transform coefficient energy feature diagnosis results,it is found that single-type feature extraction is complementary in motor faults.Therefore,a multi-feature extraction method combining Park vector distortion rate and wavelet packet transform coefficient energy features are adopted.Due to the parameter optimization problem of the traditional support vector machine classifier,in this paper,the Whale Optimization Algorithm(WOA)is applied to the parameter optimization of the Least Squares Support Vector Machine(LSS VM),and the WOA-LSS VM fault diagnosis model is established.The simulation analysis shows that the diagnosis results are better than the existing LSSVM and Particle Swarm Optimization(PSO)LSSVM.A fault experimental platform for asynchronous motors is designed and built,and the stator current signals under different fault conditions in engineering practice are collected.Feature extraction and optimization algorithm diagnosis are performed on the data in the host computer,which further verifies the feasibility of the asynchronous motor fault diagnosis technology based on multi-feature extraction and WOA-LSSVM.