Non-linear modeling and dynamic simulation of 8/6 poles SRM
For the magnetization curve of switched reluctance motor (SRM) is high saturation and has the nonlinear characteristic. This paper presents a method of modeling based on BP neural network optimized by genetic algorithm (GA). The method adopts the simple BP neural network structure based on the characteristics of flux and torque, and the network learning algorithm combines the traditional BP neural learning algorithm with GA, that means it uses the global optimization ability of GA to correct weights and thresholds of BP network, in order to overcome shortcomings of slow convergence and easy to fall into local minimum. This paper then uses this motor model to establish the simulation model of SRD in Matlab. Simulation results show the feasibility of this modeling method. And compared with the traditional BP network modeling, this method has a strong generalization ability and higher accuracy and improves the convergence rate effectively.
genetic algorithm (GA) BP neural network SRM non-linear modeling
Xiao Li Sun Hexu Zheng Yi Dong Yan Gao Feng
School of Control Science and Engineering, Hebei University of Technology, China, Tianjin 300130 School of Control Science and Engineering, Hebei University of Technology, China, Tianjin 300130 Sch
国际会议
The 31st Chinese Control Conference(第三十一届中国控制会议)
合肥
英文
4534-4538
2012-07-01（万方平台首次上网日期，不代表论文的发表时间）