基于RBF神经网络的开关磁阻电机在线建模 及其实验验证

 为了获取开关磁阻电机(SRM)的精确模型,提出了一种基于径向基函数(RBF)神经网络对SRM进行建模的方法,主要包括离线建模和在线建模两部分。离线建模通过实验测量得到SRM的磁链特性曲线,并利用该数据训练RBF神经网络,实现SRM磁链的离线建模;在线建模是指当SRM的运行状态发生变化时,离线模型的估计磁链与实际磁链会产生误差,通过对神经网络的输出权值进行在线调节,实现具有在线动态调节功能的SRM在线模型。为了验证该方法的可行性,针对一台12/8结构的SRM样机进行仿真和实验,结果表明SRM的离线模型和在线模型在仿真和实验条件下均能正确地估计SRM的磁链特性,而且在线模型的估计精度高于离线模型,验证了本文的研究方法合理有效。

Online Modeling for Switched Reluctance Motor Using Radial Basis Function Neural Network and Its Experimental Validation

To obtain the accurate switched reluctance motor(SRM) model,the offline and online modeling schemes based on radial basis fuction(RBF) neural network are investigated in this paper.Firstly,an offline modeling scheme is studied.The flux linkage characteristics are obtained from experiment and used as a training data set,based on which an RBF neural network is trained to obtain the offline SRM model.Secondly,an online modeling method is proposed.When the operating conditions of the SRM changes,the offline model is not able to approximate the real-time SRM characteristics accurately.Based on the approximation error,an online modeling scheme is applied to improve the model accuracy by regulating the values of the RBF weights online.To verify the feasibility of this method,both the offline and online modeling schemes are tested in simulations and experiments using a 12/8 SRM.The results show that both the offline and online models can estimate the flux linkage characteristics correctly and the online model is more accurate than the offline model.Simulation and experimental results have verified the effectiveness and advantages of this modeling methods.