Research On Speed Control And Fault State Perception Method Based On Six-phase PMSM

With the increasingly serious air pollution and greenhouse effect,the emission from traditional fossil fuel vehicle is regarded as the important reason.Electric vehicle(EV)has the characteristics of non-pollution,clean and environment protection because it uses unconventional fuel as the power which has been paid more attention.Permanent magnet synchronous motor(PMSM)has been used on EV widely with the advantages of high efficiency,high torque density and multiple control freedom.Moreover,compared with traditional three phase PMSM,multi-phase PMSM has the high power level and ability of fault tolerant operation.It could be foreseen that multi-phase PMSM would have a vast potential for future EV development.In order to avoid the more serious motor fault and system breakdown which were caused by open phase during multi-phase PMSM running,it is necessary to do the fault prediction and diagnosis.In this paper,the major research points are the six-phase PMSM speed control and fault state perception method.In the research of six-phase PMSM speed control,the model of neutral point isolation six-phase PMSM shifted by 30° has been built by vector space decoupling.The precision of the model has been verified with MATLAB own three-phase PMSM model,as the same time,it has also been compared with experiment six-phase PMSM.The vector control is used to design the rotate speed-current dual closed-loop control system.In order to realize the good performance of rotate speed tracking,a variable-structure sliding mode control(SMC)was designed and compared with traditional PI controller to verify that this strategy has the characteristics of rapid response,no overshoot and robustness.In the research of fault state perception method for six-phase PMSM,this paper derived mathematical model of neutral point isolation six-phase shifted by 30° with principle of stator magnetomotive force invariance.Wavelet packet analysis was used to collect the feature values and wavelet neural network was built to do the fault prediction and avoid system spurious triggering.The wavelet K-Nearest Neighbor(KNN)machine learning system has also been built to diagnose the fault types quickly which could realize fault state perception.MATLAB and Scikit-Learn library of Python were used to do simulations which could verify the strategy reliable and effective.Last but not the least,the experiment platform has been set up with dSPACE controller and six-phase PMSM.Using speed control experiment to verify the precision of mathematical model and the control strategy could realize the rapid tracking with robustness.Meanwhile,it could also provide the basic of later experiments of fault diagnosis and fault-tolerant control.