The Study On The Optimization For The Structural Parameters Of Permanent Magnet Synchronous Motor Based On Ansoft

The deepening study on permanent magnet synchronization motor(PMSM) promotes the development of motor optimized design. It also sets higher requirements for the optimization algorithm. The structural diversity of PMSM increases the complexity of the internal magnetic field and causes equivalent magnetic method and other traditional methods cannot achieve the needed accuracy. Although the numerical analysis of electromagnetic field has a good accuracy, the consumption of its calculation is too large. Therefore, a new algorithm is needed to shorten the cycle of motor design.This article starts with the determination of basic dimensions and the traditional performance analysis method, after introducing the relative theories of two-dimensional finite element method(FEM), then is extended to numerical analysis method of electromagnetic field. On the basis, builds the initial model of motor by the software for motor design analysis and analyses and calculates the performance of motor. To improve the performance of motor, selects the thickness of pole, pole-arc coefficient, polar arcs eccentricity and the length of air gap as the design variables; the cogging torque and the sinusoidal distortion rate of no-load air-gap flux density waveform as the objective functions for optimization. At first, determines the ranges of the variables’ values through the simulation experiments, then designs orthogonal text to obtain the sample space needed by regression analysis. After that, builds the response surface model for each objective function based on the support vector machine(SVM) separately. Uses the particle swarm optimization(PSO) with mutation operation to achieve the single-objective optimization of two objective functions separately. Substitutes the optimized results into the FEM software, the cogging torque falls to 0.264N?m from 4.37N?m, the sinusoidal distortion rate of no-load air-gap flux density waveform falls to 17.36% from 29.14%, the simulation experiments verifies the accuracy of the results. At last, in view of the actual optimization of motor is always the multi-objective optimization problem, puts the two objective functions in a process of optimization to achieve the multi-objective by PSO at the same time. After simulation the cogging torque becomes 0.31N?m and the the sinusoidal distortion rate of no-load air-gap flux density waveform becomes 22.33%, both of them are ideal.The optimized results of “SVM+PSO” algorithm is good and can ensure a high degree of accuracy. At the same time, the needed sample space is small, and it can finish the optimization after fewer evolutions because of its fast convergence speed, the two algorithms continue their own advances after the combination. Moreover, support vector regression uses the “black box method”, reduces people’s dependence on motor knowledge significantly and simplifies the motor design process effectively. As the generality of the algorithm, it also provides the guidance and reference for the optimization of motor’s other performance parameters.