基于在线模糊神经网络建模的开关磁阻电机高性能转矩控制

摘要

针对开关磁阻电机(SRM)的转矩脉动问题,提出了一种新的SRM转矩控制方案.首先应用自适应模糊神经网络(ANFIS)对SRM静态转矩逆模型和磁链模型进行离线学习,然后根据转矩分配函数对各相转矩进行分配,利用ANFIS转矩逆模型求出期望转矩下的SRM优化相电流波形.考虑到离线模型的局限性和实时运行时电机中存在的参数变化等不确定因素,通过在线监督学习的方法调整ANFIS转矩逆模型和磁链模璎的参数以提高模型的准确性.基于在线调整的ANFIS磁链模型设计自适应滑模控制器调节SRM相绕组中的实际电流跟踪期望相电流波形,从而实现其高性能转矩控制.%A novel torque control scheme for switched reluctance motor( SRM ) to reduce its torque ripple was proposed.Firstly, an adaptive neural fuzzy inference system(ANFIS) was designed to learn the nonlinear static position-torque-current characteristic and the flux-linkage characteristic of an SRM offline.Then each phase torque was calculated according to torque share function and the desired phase current waveform was obtained using the ANFIS inverse torque model.Considering the limitation of the offline model and the uncertainties existing in the real-time motor system, the parameters of ANFIS was turned through online supervised learning to improve the accuracy of the inverse torque and the flux-linkage model.Based on the online flux-linkage model, an adaptive sliding-mode current controller was designed to regulate the actual SRM phase winding current to track the desired phase current waveform,thereby reduce the torque ripple of SRM.