电流自适应控制抑制开关磁阻电机转矩脉动

开关磁阻电机(SRM)的强非线性源自其双凸极结构、磁路非线性和脉冲供电方式。传统控制多采用SRM线性转矩模型求得参考电流,导致其运行时转矩脉动大。提出基于转矩偏差的双权值神经网络(DWNN)自适应PID控制与基于有限差分扩展卡尔曼滤波(FDEKF)预测电流的前馈补偿控制相结合的SRM控制策略。(1)加入偏差预处理,对转矩偏差进行非线性处理,实现"小误差,大增益,大误差,小增益"的控制,以此为基础进行双权值神经网络自适应PID的电流控制;(2)采用预测电流,构成参考电流的前馈补偿控制,提高控制系统一步预测能力。基于有限差分扩展卡尔曼滤波预测电流,将其与参考电流之差实时补偿参考电流,优化得到恒转矩下有效的控制电流,间接实现总转矩的有效控制。仿真结果证明所提控制策略能有效抑制SRM的转矩脉动。

Torque ripple suppression for switched reluctance motors based on current self-adaptive control

A switched reluctance motor (SRM) has strong nonlinear characteristics due to its double-salient structure, nonlinear magnetic circuit and pulse power supply mode. In traditional control, the SRM linear torque model is used to calculate reference current to lead to large torque ripple when operating. Here, the SRM control strategy combining the self-adaptive PID control based on torque deviation’s double-weight neural network (DWNN) and the feed-forward compensation control based on finite-difference extended Kalman filtering (FDEKF) to predict current was proposed. The pretreatment of deviation was used to nonlinearly process torque deviation to realize the control with "small error, large gain, large error, small gain". Then, the self-adaptive PID control based on DWNN was used to control current. The current prediction was adopted to form the reference current’s feed-forward compensation control to improve the one-step predictive ability of the control system. The current was predicted based on FDEKF, the difference between the predicted current and the reference one was used to compensate the reference current in real time. After optimization, the effective controlled current was obtained under constant torque to realize indirectly the effective control of the total torque. Simulation results showed that the proposed control strategy can effectively suppress torque ripple of SRMs.