一种改进的RBF神经网络的开关磁阻电机磁链模型  

An Improved Flux Linkage Model of Switched Reluctance Motors Based on RBF Neural Networks

作　　者：胡学彬[1] 王宏华[1] 周湛权 

机构地区：[1]河海大学,南京江苏211100

出　　处：《机械制造与自动化》2015年第6期172-175,共4页Machine Building & Automation

摘　　要：基于径向基(RBF)神经网络的开关磁阻电机无位置传感器技术在拟合磁链-电流-角度的模型中,初始角度范围内因为磁链的变化率较小,数据点密集,RBF神经网络的聚类中心点难以准确地计算,导致拟合的误差比较大。采用分段的形式对RBF神经网络的开关磁阻电机无位置传感器模型进行改进,即将RBF神经网络模型的小角度范围内的数据分离出来予以单独处理,用线性化的方法对其进行建模通过MATLAB仿真验证了该方法的可行性。In the process of fitting flux linkage/current/angle in the sensorless techniques of SRM based on RBF neural networks, at a small range of initial angle the rate of change is so small and the sample point is so closed that it is hard to compute the cluster center RBF neural network correctly and the error in fitting is caused. This paper improves the flux linkage/current/angle in the sensorless techniques of SRM by way of piecewise made, it means that the sample data is separated and handled in the small range of the initial angle, then the linearization technique is used to establish the model. It proves the method is available by MATLAB simulation.

关 键 词：开关磁阻电机 无位置传感器 径向基神经网络 

分 类 号：TM352[电气工程—电机]