一种基于模糊神经网络和遗传算法的智能PID控制器

An intelligent PID controller based on fuzzy neural network and genetic algorithm

常规的PID控制器参数整定方法需要被控对象的精确数学模型,且整定出的参数不能进行在线调整.而模糊控制和神经网络均不依赖被控对象的数学模型,且具有较强的自适应和自学习能力;遗传算法则是一种新型的全局优化方法.鉴于此,提出将模糊控制、神经网络和遗传算法引入PID控制器的设计过程.首先,运用遗传算法优化隶属度函数的中心值和宽度,并借助模糊逻辑控制确定遗传算法中的交叉概率和变异概率.然后,再运用BP算法优化模糊神经网络的连接权系数.仿真结果表明,该方法提高了系统的自适应能力和抗干扰能力,增强了系统的鲁棒性.

Traditional methods of designing the parameters of PID controller need a precise mathematic model of object. And the controller can not adjust itself on line to the variation of the surroundings and the object. In contrast, a fuzzy control and neural network need not have any mathematic model of object, and have fine adapt ability and learning ability. Genetic algorithm is a new global optimization one. So it was proposed that the fuzzy control, neural network, and genetic algorithm be introduced into the designing process of PID controllers. Fist, the genetic algorithm was used to optimize the central value and width of the membership function. And at the same time, the fuzzy logic control was lent to determine crossover and mutation probability in the genetic algorithm. Finally, the back propagation algorithm was used to optimize the connection coefficients of fuzzy neural network. The simulation result showed that this method improved the adapt ability, anti-interference, and robustness of the system.