数据驱动的无模型自适应直线伺服系统精密控制和实现

针对现代制造业对高精度机床伺服系统的要求, 将数据驱动的无模型自适应控制方法应用到直线伺服系统的位置控制中, 控制器设计不包括直线伺服系统结构的任何信息, 是直接基于动态线性化模型中伪偏导数的估计和预报, 而伪偏导数是根据直线电机电压输入和位置输出在线估计的. 永磁同步直线电机运动控制系统的实时实验结果表明, 在相同条件下, 数据驱动的无模型自适应控制方法的位置跟踪误差比PID减小了0.4mm到2.6 mm, 比神经网络控制时减小了0.2mm到0.5 mm. 该方法还提高了对负载扰动的鲁棒性.

Data-driven model-free adaptive precision control for linear servo system

To meet the requirements on the high-precision tool-servo system in modern manufacturing industry, we propose a data-driven model-free adaptive control (MFAC) method to control the position of this system. The design of controller requires no structure information of the linear servo system, but makes use of the estimated and predicted pseudo-partial derivatives (PPD) of the dynamic linearization model, that are obtained online from the input-voltage and the output-position of the linear motor. Under the same conditions in real experiments, this method brings about a positiontracking error which is 0.4mm to 2.6mm smaller than that of the PID method, and 0.2mm to 0.5mm smaller than that of the neural networks (NN) method. The robustness against the load disturbance is also improved.