基于BP神经网络的三轴增稳云台自抗扰控制

Active Disturbance Rejection Control of Three-Axis Stabilized Platform Based on BP Neural Network

针对三轴增稳云台伺服系统非线性特性,以及PD控制抗扰能力差,自抗扰控制器由于参数众多而导致整定过程耗时且费力的缺陷,本文利用BP神经网络的全局逼近能力和自我学习能力,将其与自抗扰控制器组成复合控制器,对自抗扰控制器的所有关键参数进行自整定寻优,应用于含Stribeck摩擦模型的三轴增稳云台伺服系统。仿真结果表明:该方法用于自动整定参数可行有效,与PD控制和参数固定的常规自抗扰控制器相比,具有更高的控制精度和更强的抗扰能力,对提高增稳云台的性能具有较好的应用价值。

In view of the non-linear characteristics of the three-axis stabilized pan-tilt servo system, the anti-disturbance ability of PD control is poor, and the setting process of the active disturbance rejection control is time-consuming and laborious due to the large number of parameters. By using the global approximation ability and self-learning ability of BP neural network, a composite controller is composed of BP neural network and active disturbance rejection control. All the key parameters of active disturbance rejection control are self-tuned and optimized, which is applied to the three-axis stabilized pan-tilt servo system with Stribeck friction model. The simulation results show that the method is feasible and effective for parameter auto-tuning. Compared with the conventional ADRC with fixed parameters and PD control, it has higher control accuracy and stronger anti-disturbance ability, and has better application value for improving the performance of the stabilized platform.