Research On Fault Diagnosis Method Of Permanent Magnet Synchronous Motor Based On Sparse Representation

Since its birth,the permanent magnet synchronous motor has been increasingly used in industrial fields.However,due to the complex and changing industrial environment,affected by the power supply and load conditions,the motor will inevitably have various failures during long-term continuous operation.Therefore,it is very important to study the fault detection and diagnosis technology of permanent magnet synchronous motors.Among various fault diagnosis techniques,this paper studies emerging diagnostic techniques based on sparse representation theory,including using sparse representation to extract fault features,to classify faults,and combining it with support vector machines.At the same time,a motor fault diagnosis research platform was developed using LabVIEW,and experiments were performed based on the platform using real data collected from permanent magnet synchronous motors to verify the effectiveness of the proposed diagnostic method.In this paper,the research background and significance of the fault diagnosis of permanent magnet synchronous motors are first introduced,the common faults of permanent magnet synchronous motors are analyzed,and the research status of motor fault diagnosis technology is summarized,thus leading to the sparse representation used in this work.The research contents of this paper are as follows:(1)The fault feature extraction method based on sparse representation is studied.After introducing the basic principles of sparse representation theory,the method of fault feature extraction using sparse representation is studied,which combines current and vibration signals.This method uses orthogonal matching pursuit algorithm to solve,and uses support vector machine to classify the extracted features.At the same time,the construction and application of the motor condition monitoring and fault diagnosis experimental platform are introduced,and experiments are performed based on the signals extracted from the actual inter-turn short circuit fault motor.(2)The application of sparse representation in intelligent fault diagnosis is studied.After introducing the basic principles of sparse representation-based classification and dictionary learning algorithms,a fault diagnosis method based on sparse representation classification is proposed.This method diagnoses faults according to the feature sample set extracted by wavelet packet analysis,and uses sparse representation classification with K-SVD for identification.After that,the actual motor data of multiple fault types were collected through the experimental platform,and the method was verified with it.(3)A fault diagnosis method combining sparse representation and support vector machine is studied.After introducing the multi-classification algorithm of support vector machine,a joint fault diagnosis model combining sparse representation classification and support vector machine is proposed.It is implemented by calculating the variance of residual distribution in sparse representation classification and setting the variance threshold.Multi-fault diagnosis experiments on actual permanent magnet synchronous motors prove the effectiveness and accuracy of the proposed model.