为了实现开关磁阻电机调速系统(SRD)的准确动态仿真,在测取准确磁特性样本数据基础上,利用神经网络所具有的非线性映射能力,采用基于Levenberg-Marqvardt算法的BP神经网络,建立了开关磁阻电动机的非线性模型,并在MATLAB仿真平台上搭建SRD系统动态仿真模型。仿真实验表明,与常规线性SRD动态仿真模型相比,采用BP神经网络的SRD动态仿真模型转矩脉动小,具有稳定性好,鲁棒性强的特点。
To provide the accurate dynamic simulation of the Switched Reluctance Driving( SRD), After measuring the accurate flux-linkage data, the nonlinear model of Switched Reluctance Motor(SRM) was founded, which made use of nonlinear mapping ability of BP neural network based on Levenberg - Mar- quardt arithmetic. Based on the BP neural net nonlinear model, a complete simulation model of SRD on MATLAB was developed. Simulation experiment indicated that the SRD simulation model based on the BP neural net has smaller fluctuations of torque and performs better than the linear model of SRD on the stability and robustness.