某电液伺服系统的神经网络自抗扰控制

Active Disturbances Rejection Control of Certain Electro-hydraulic Servo System Based on Neural Network

针对某电液伺服系统存在非线性及时变性,难以对其进行精确控制的问题,提出了神经网络自抗扰控制方案。该方案利用神经网络所具有的可以实现任意复杂映射关系的能力,将非线性误差反馈控制规律中的比例系数与微分增益作为单神经元自适应控制器的权系数,通过单神经元的自学习功能进行在线调节;同时利用RBF神经网络作为辨识器,以辨识被控对象的梯度信息。通过MATLAB计算机仿真结果证明,该控制方案使系统具有响应速度快、稳态精度高、鲁棒性强等优点,并能够有效的抑制外界扰动。

In order to ensure the static and dynamic performance, an active disturbances rejection controller was proposed to control certain electro-hydraulic servo system, which presents nonlinear and time-variation. Based on the ability that the neural networks can come true arbitrary complex mapping relations and the nonlinear error feedback control law, the single neuron adaptive controller introduces proportionality coefficient and differential gain as weight coefficients. The learning of single neuron is online-adjust .RBF can identify gradient information of this system. The MATLAB simulation results show that the system have advantages of speedy response, high steady-state accuracy and strong robustness against modeling uncertainty and external disturbance.