基于预处理BP神经网络的开关磁阻电机建模

针对开关磁阻电机(SRM)强耦合、强非线性、难以精确解析建模的问题,提出一种基于数据预处理的反向传播(BP)神经网络建模方法。首先通过传统直流脉冲法测量一个电周期内SRM静态电磁特性,获取建模样本数据;其次充分利用电机先验知识,通过可以初步反映SRM非线性特性的磁链和转矩解析表达式对实测样本数据进行预处理并作为BP神经网络新的输入,降低神经网络拟合误差。与传统BP神经网络建模的对比结果显示,引入预处理方法可以有效减少BP神经网络节点数量,增强神经网络泛化能力,提高神经网络建模精度。

Modeling of Switched Reluctance Motor Based onPretreatment BP Neural Network

Aiming at the problem that the switched reluctance motor (SRM) with strong coupling and strong nonlinearity was difficult to accurately resolve and model, a backpropagation (BP) neural network modeling method based on data pretreatment was proposed. Firstly, the static electromagnetic characteristics of SRM in one electrical cycle were measured by the traditional DC pulse method to obtain modeling sample data. Secondly, the motor prior knowledge was fully utilized, and the measured sample data were preprocessed through the flux linkage and torque analytical expressions which could initially reflect the nonlinear characteristics of the SRM, and then sent to the BP neural network so as to reduce the neural network fit error. Compared with the traditional BP neural network modeling, the pretreatment method could effectively reduce the number of BP neural network nodes, enhance the generalization ability of the neural network, and improve the modeling accuracy of the neural network.