【作者】 扈宏杰； 孟召赞；

【Author】 HU Hong-jie, MENG Zhao-zan (School of Automation Science and Electrical Engineering, Beihang University, Beijing 100083, China)

【机构】 北京航空航天大学自动化科学与电气工程学院；

【摘要】 经典的、基于对象模型的PI控制方法简单、易于实现,但对于一些负载扰动和模型参数的变化,往往不能起到很好的抑制作用。针对上述问题,提出了一种神经网络自适应PI的控制方法,利用负载干扰观测器和神经网络自适应地调整PI控制器的参数,从而来有效地减少负载的干扰和模型参数的变化对系统造成的影响,提高了系统的鲁棒性。仿真结果表明了该方法的有效性。更多还原

【Abstract】 The traditional PI control method based on a plant model is simple, and easy to realize. But, it has not a good restraining effect on the varieties of load disturbance and model parameters. To resolve the problems, a self-tuning PI control method was proposed using a neural network. The PI parameters are adjusted by use of the load disturbance observer and neural network identifier, so the influence caused by the load disturbance and model parameters varieties are weakened, and the self-tuning PI control method improves the system robustness. The simulation result shows that the method is valid.更多还原