介绍了在理想状态下利用电磁场和电路的基本理论在四层线性模型的基础上所建立的PMLSM电路模型以及利用BP算法用C语言编制程序实现一个单输入层单输出层 4隐含层的神经网络 ,然后用试验数据对网络进行学习训练而建立的PMLSM网络模型 .以一台单边型PMLSM为例比较了这两种模型 Introduced in the ideal state of the electromagnetic field and the use of the basic theory of the circuit in the four-tier linear model based on the established PMLSM circuit model and the use of BP algorithm C programming language to achieve a single input layer single output layer 4 hidden layer of nerve Network, and then use the experimental data to train the network to build the PMLSM network model.Using a unilateral PMLSM as an example to compare the two models