基于神经网络干扰观测器的巷道机器人鲁棒控制方法研究

Research on Robust Control Method of Roadway Robot Based on Neural Network Interference Observer

以巷道机器人铲斗油缸电液控制系统为研究对象,提出了一种基于神经网络干扰观测器的鲁棒控制器来提高强时变负载力工况下油缸的控制水平。该控制器利用神经网络的学习功能,实时估计和补偿未知负载力。采用反步法设计了鲁棒控制器并进行了系统稳定性的验证。通过Simulink/Simscape搭建了铲斗油缸系统的物理模型,并进行了仿真分析。神经网络干扰观测器可有效完成对未知负载力的实时估计,经过负载力补偿后的鲁棒控制器相对PID控制器精度提高了约79%。

Electro-hydraulic control system of bucket cylinder of roadway robot was taken as research object,and a robust controller based on neural network interference observer was proposed to improve the control level of cylinder under strong time-varying load conditions.The controller uses the learning function of neural network to estimate and compensate the unknown load in real time.The robust controller was designed by backstepping method and the stability of the system was proved.The physical model of bucket cylinder system was built by Simulink/Simscape,and the simulation analysis was carried out.The neural network interference observer can effectively estimate the unknown load in real time,and the accuracy of the robust controller after load compensation is improved by about 79%compared with PID controller.