基于智能滑模观测器的PMLSM调速系统研究

Research of Speed Regulating System Based on Intelligent SMO for PMLSM

针对由于传统滑模观测器存在而引起系统抖振较大的问题,设计了一种可在线学习BP神经网络滑模观测器,以减小系统抖振和提高永磁直线电机伺服控制系统的性能。通过设计滑模观测器进行电流估计,获得反电势大小;将BP神经网络与传统滑模观测器相结合,并将电机定子电流估计值与实测值间的误差作为性能指标函数,实现权值的在线学习,达到滑模观测器增益参数最优化自整定目的;引入锁相环技术达到对电机动子位置和速度的辨识。仿真实验结果表明,基于BP神经网络的滑模观测器能够实现对电机动子位置和速度的准确观测,且系统响应快速,稳态精度高。

For reducing the chattering phenomenon generated by the conventional sliding mode observer（SMO）,a new SMO based on the BP neural network was designed to improve the servo control system performance of the permanent magnet linear synchronous motor（PMLSM）. A SMO was designed to estimate the stator currents to obtain the back EMF. The BP neural network was introduced into the SMO to realize the self-optimization of the SMO＇s gain parameters,the error between the estimated current and measured current as the performance index function to realize the on-line weight learning. The phase locked loop technology（PLL）was applied to estimate the mover velocity and position. The simulation results show the new SMO may accurately observe the mover position and speed,and the control system has quick response and high stability accuracy.