Research On Inter-Turn Fault Diagnosis Of High-Speed Permanent Magnet Synchronous Motor

High Speed Permanent Magnet Synchronous Motors(HSPMSMs)have significant advantages such as high power density and high efficiency,and have been used more and more widely.The safety and reliability requirements of HSPMSMs and their systems are very high in applications such as fluid machinery,electric spindles,and electrified transportation.The most common fault of HSPMSM is the inter-turn short circuit of stator winding.If the fault cannot be detected and eliminated in time,it can easily expand or even burn the motor,damage the electromechanical system,and cause major economic losses and serious safety hazards.Therefore,this paper takes HSPMSM as the main research object,conducting related researches on the establishment of motor fault model,the selection of fault-related feature quantities,and the effective fault diagnosis method.The existing diagnostic methods for motor inter-turn short-circuit faults(ITSCFs)are combed and their advantages and disadvantages are analyzed in this paper.Aiming at the characteristics of many parallel windings of HSPMSM windings,the ITSCF model considering the series or parallel structure of the windings and the ITSCF model considering the parallel conductors of the coil are established.The voltage equation and the short circuit equation in the dq coordinate system of the motor after an ITSCF are further deduced.And the motor models of different degrees of fineness are unified in structural form.The proposed motor model is built in MATLAB/Simulink.Simulations are carried out under the condition of motor health status and different severities and different types of inter-turn faults.Results are compared with the output of the finite element model to verify the validity and accuracy of the proposed model.The results indicate that the proposed model can not only reflect the characteristics of the faulty motor,but also has a fast simulation speed.Furthermore,it is easy to modify the fault mode and set the severity of fault.Thus large numbers of simulation data of the motor in different operating states are obtained.In this paper,the current and voltage characteristic quantities related to the ITSCF are analyzed.And the changing trend of each characteristic quantity with the severity of the motor fault and the speed of the motor is summarized.Besides,the RMS,fundamental and third harmonic amplitudes of the three-phase current,together with the fundamental and second harmonic amplitudes of the dqaxis control voltage are selected as the fault characteristic quantities and the input of the diagnostic algorithm proposed in this paper.Deep learning algorithm combined with machine learning algorithm is chosen for fault diagnosis and classification.A diagnosis algorithm based on sparse auto-encoder with least squares support vector machine is proposed,which not only ensures the computational efficiency,but also takes into account the adaptability of the model.The training results based on simulated datasets show that the proposed algorithm has higher diagnostic accuracy and sample adaptability than traditional classifiers.Finally,the paper takes a 12 kw,1250Hz high-speed motor designed and manufactured as the experimental object.The windings of the motor are modified to simulate the ITSCF and the insulation failures between different parallel conductors.And a fault test platform is built.Based on Raspberry Pi 3B,a fault diagnosis module for motor turn-to-turn short circuit is implemented.Experiments are carried out in the motor health state and different fault states,and online diagnosis is carried out in the fault diagnosis module,which verifies the effectiveness of the proposed diagnosis algorithm.