针对永磁直线同步电机(permanent magnet linear synchronous machine,PMLSM)有限元仿真模型的计算时间长,不能直观地显示结构参数与输出推力的关系,无法指导电机结构参数优化等问题,提出基于子域解析法和深度神经网络算法的PMLSM改进仿真模型,根据麦克斯韦方程组计算得到电机的磁通密度、空载反电势等性能数据,结合深度神经网络算法拟合出电机结构参数与输出推力的非线性关系.基于此模型,使用自适应遗传算法对PMLSM的推力密度进行优化,并与有限元仿真结果对比.结果表明:PMLSM改进仿真模型的计算速度是有限元模型的87.1倍,推力计算结果与有限元结果的平均误差为2.87%,优化后的电机推力密度提高了5.7%.
Simulation and Optimization of Permanent Magnet Linear Machine Based on Deep Neural Network
The finite element model(FEM)of permanent magnet linear synchronous machines(PMLSMs)takes a long computing time and cannot directly display the relationship between structural parameters and output thrust,thus failing to guide the structural parameter optimization of the machine.An improved simulation model of PMLSMs based on the subdomain analytical method and deep neural network(DNN)algorithm is proposed.The magnetic flux density,no-load counter electromotive force(EMF),and other data are obtained according to Maxwell's equations.The nonlinear relationship between the structural parameters of the machine and output thrust is fitted by the DNN algorithm.Based on this model,an adaptive genetic algorithm is used to optimize the thrust density of PMLSMs,and the results are compared with the finite element simulation.The results show that the computing speed of the improved simulation model of PMLSMs is 87.1 times that of the FEM.The average error of thrust between these two models is 2.87%,and the optimized thrust density of the machine is increased by 5.7%.
万方数据
维普
