基于神经网络推力观测器的永磁直线电机滑模控制

针对永磁直线同步电机(PMLSM)伺服系统易受到负载扰动、参数变化和推力波动的问题,为确保在较宽的速度范围内实现更为精确的速度控制,采用基于神经网络推力观测器的滑模控制取代常规的PI控制,滑模变结构控制律采用等效控制法,加上负载扰动前馈补偿项,而扰动补偿则是通过线性推力观测器并联一个神经网络观测器相加而得.实验和仿真结果表明,此方法较常规PI控制提高了跟踪性能,增强了伺服系统对参数摄动和外在扰动的稳健性.

Sliding Mode Control Bused on Neural Network Thrust Observer for PMLSM

Aiming at the permanent magnet linear synchronous motor servo system is vulnerable to load disturbance, parameters change and thrust fluctuations, the sliding mode control based on neural network thrust observer instead of the regular proportion integration(PI) was indicated to ensure that it achieves a more precise speed control with a wide range of speed. Sliding mode control regular adopts the equivalent control way, together with the load disturbance feed-forward compensation, and it is consist of linear observer and neural network observer. Experiment and simulation results show that it can improve the tracking perform-ance, and enhance the stability and robustness of parameter perturbation and external disturbances.