基于BP神经网络的分数阶P IαDβ控制器参数整定研究

摘要

Fractional order PIαDβcontroller is more flexible and offers opportunity to obtain better control quality of a fractional order control system than the traditional PID controller. However, the parameters tuning of fractional order controller is more complex. Aming at the problem, a method of fractional order PIαDβ controller tuning based on BP neural network is presented, which efficiently overcome the complexity of fractional order controller tuning. A fractional order PIαDβcontroller and an integer order PID controller are tuned respectively by using the proposed method. Results show that fractional order PIαDβcontroller can get better performance than inte-ger order PID controller.%对于分数阶控制系统，因分数阶控制器较传统PID控制器具有更好的适应性，这就为得到更加细腻的控制品质提供了可能。然而，分数阶PIαDβ控制器的参数整定则相对复杂。针对这一问题，提出了基于BP神经网络的分数阶控制器参数整定方法，有效地克服了分数阶控制器参数整定复杂问题。采用文章所提方法，分别设计了整数阶PID、分数阶PIαDβ控制器，并进行仿真对比。结果表明，采用分数阶PIαDβ控制器的系统控制品质优于整数阶PID控制器。