混沌粒子群优化的模糊神经PID控制器设计

Design of the fuzzy neural PID controller based on hybrid PSO

针对常规PID控制的线性局限性及传统模糊控制和模糊PID控制中积分误差规则难以获取,系统存在稳态误差的问题,提出一类以模糊神经网络和PID神经网络组成的模糊神经PID控制器.以整个神经网络的权值为优化参数,利用基于混沌策略的粒子群全局优化算法离线优化和误差反传算法在线调整相结合的方法获得控制器参数,并设计了混沌优化与粒子群结合的两步方案.仿真结果表明:与传统PID、模糊、模糊PID控制相比,系统的瞬态和稳态性能有了明显提高,且保持了一定的鲁棒性及高跟踪精度.该方法有效地拓展了PID控制的使用范围,并为智能方法与PID控制的结合提供了一种新的参考方案.

Considering the problems of the linearity limit of PID control and the steady-state error in fuzzy, fuzzy PID control for it cannot easily obtain the control rules of the integral error, so a fuzzy neural PID controller which consists of a fuzzy neural network and a PID neural network is designed. The parameters of the controller are optimized by the mixed learning methods integrating the offline particle swarm optimization algorithm combined with chaos strategies of global searching ability, with the online BP algorithm of local searching ability. Simulation results show that the designed novel controller and the proposed optimization algorithm have obviously improved the performance of the transient state and steady state in the control processing. Compared with conventional PID, fuzzy control and fuzzy-PID control method, the new controller with the optimization method has good robustness and better performance. The new method breaks through the limit of linearity of PID control and expands its applications. It also provides a new reference for the combination of intelligent method and PID method.