开关磁阻电机神经网络随控无位置传感器实现

Control Configured Position Sensorless Achievement of Switched Reluctance Motor Using Neural Network

采用以开关磁阻电机(SRM)控制方式与无位置传感器控制相结合的方法,实现无位置传感器控制(即随控无位置传感器)。对SRM采用直接转矩控制(DTC),并在此基础上推导出以绕组相电流和相电压为输入,以各相转矩为输出的非线性映射关系。利用广义回归神经网络的非线性拟合能力实现该输入、输出之间的非线性映射,进行转矩估计,从而消去转矩传感器。仿真结果表明,SRM不仅可以在各种速度下实现无位置传感器控制,还可以有效的抑制转矩脉动。

This paper presented an approach that the control strategy of switched reluctance motors （SRM） combined with the methof of position sensorless control to achieve the position sensorless control （as is socalled control configured position sensorless ）. A new method that combines the generalized regression neural networks（GRNN） with direct torque control （DTC） was proposed to accomplish the positon sensorless control of SRM. The nonlinear mapping relation that winding phase current and winding phase voltage as input and torque as output was elicited after the control strategy of DTC was adopted by SRM. Nonlinear mapping between input and output was completed through using the ability of the nonlinear mapping of GRNN in order to estimate the magnitude of torque and remove torque sensor. The simulation shows that this method can achieve the positon sensorless control of SRM at all kinds of speed, and restrain torque ripple effectively.