基于自适应灰色预测和干扰观测器的伺服干扰抑制方法

Method of Servo Interference Suppression Based on Adaptive Grey Predictive Controller and Disturbance Observer

为提高稳定平台伺服系统的响应和抗干扰能力,提出了一种基于自适应灰色预测（AGPC）——分数阶改进干扰观测器（FIDOB）的稳定平台伺服干扰抑制方法。GM（1,1）幂模型对系统输出进行建模并设计了自调节模块,将预测误差和实际误差加权合成一个综合误差,分别根据实际误差和预测误差的大小同时调节预测步长和预测误差的权值,提高系统的响应性,减小预测误差对系统的输出影响;构造了分数阶改进干扰观测器,并详细推导了分数阶改进干扰观测器的鲁棒稳定性。最后通过数值仿真实验表明,该方法不仅可以有效抑制稳定平台外界干扰和测量噪声,而且提高了系统响应能力。仿真实验中,在摩擦和测量噪声干扰情况下,稳定平台系统速度环的跟踪误差可以达到不超过0.1 rad/s。在静态和动态实验中,稳定平台的调节时间缩短了0.258 s,稳定精度提高了约1.5°-2.5°。

In order to improve the response and vibration suppression abilities of stabilized platform servo system, a novel vibration suppression strategy was proposed by using an integration of fractional-order improved disturbance observer （FIDOB） based on adaptive grey predictive controller （AGPC） in stabilized platform system. A method of system modeling output module was established based on GM （1,1） power module and an adaptive adjust module was designed. The predicted error and actual error were combined together to form an integrate error, according to the values of the actual control system error and the predicted error, the prediction step and predicted error weight were adjusted to improve the response ability of system and decrease the influence of predicted error on the system output. The fractional-order improved DOB was established, and the robust stability was derived in detail. FIDOB was used to obtain disturbance estimate and generate compensation signal, and as the order of Q-filter was expanded to real- number domain, FIDOB had a wide range to select a suitable tradeoff between robustness and vibration suppression. Finally, numerical simulation results illustrated that the method can suppress external disturbances and measurement noise well, and it can also improve the response ability of system. The proposed control strategy was simple in control-law derivation, and its effectiveness was validated by numerical simulations. In numerical simulation experiments, with the interference of friction and measurement noise, the tracking error of stabilized platform system was no more than 0. 1 rad/s. In the static and dynamic experiments, the regulating time of the stabilized platform was decreased by 0. 258 s, and the stable precision was increased by 1.5°-2.5°.