Permanent Magnet Synchronous Motor Fault Diagnosis Based On Swarm Intelligence

Nowadays,industrial system has the trend of being automatic and intelligent.Stablility of a system is fundamental to the automation and intelligence of the system.Permanent magnet synchronous motor(PMSM)has the advantage of high efficiency,high power density and strong performance of robustness.The application of permanent magnet synchronous motor has been indispensable to modern industrial,especially the field of fine control.When a fault of motor is failed to be found or processed timely,the motor will be hurt.Much worse,the fault may destroy the whole machine and cause huge economic losses.Therefore,it is necessary and meaningful to research permanent magnet synchronous motor fault diagnosis.Driving system open circuit and stator winding inter-turn short circuit are the two most usual faults.This paper applies swarm intelligence optimization algorithms to the diagnoses of driving system open circuit and stator winding inter-turn short circuit.First of all,this paper establishes the mathematic model of dq axis of PMSM.The principle of motor vector transformation is introduced as well.Then models of PMSM when driving system open circuit or inter-turn short circuit happens are analyzed respectively.And the normal algrithms of driving system open circuit and stator winding inter-turn short circuit diagnoses are introduced.Then,in order to address the common problems of lacking of phase and interturn short circuit fault,after analyzing the basic and corresponding fault model of Permanent Magnet Synchronous Motor(PMSM),an Improved Extreme Learning Machine(IELM)algorithm is proposed based on Self-adaptive SECond-order Particle Swarm Optimization(SaSECPSO).SaSECPSO employs adaptive inertia weight and cognitive coefficient with linear variation to improve the convergence rate and accuracy of SECond-order Particle Swarm Optimization(SECPSO).In addition,the recognition rate of Extreme Learning Machine(ELM)when solving the fault model of PMSM can be significantly improved by using SaSECPSO to simultaneously optimize input weights and hidden layer threshold.Extensive experiment is carried out by taking motor speed and phase current as multi-source data sample,and the results validate that IELM has a higher diagnostic accuracy than other algorithms.Finally,in order to solve the problems of common inter-turn short circuit faults,a corresponding motor fault model based on the existing basis of Permanent Magnet Synchronous Motor is established.First,the eigenvector is extracted by using energy spectrum analysis.Secondly,the penalty factor and RBF-kernel parameter of SVM are optimized by using adaptive dynamic cat swarm optimization algorithm.Then,the optimized SVM is adopted to motor fault diagnosis.Finally,the eigenvector which obtained by energy spectrum is taken as sample data to conduct simulation experiment.The results of the experiment indicate that compared with other optimization algorithms,using ADACSO to optimize SVM parameters can improve the accuracy of SVM in fault diagnosis of Permanent Magnet Synchronous Motor.