基于自适应变异DE算法的PID参数整定优化

摘要

针对工业过程中的PID参数整定难的问题,在分析传统差分进化算法的基础上,提出了一种自适应变异的差分进化算法,用于PID控制器的参数优化.改进算法定义了群体相似度系数和个体优劣系数,根据群体相似度系数动态调整变异操作,发挥不同变异操作模式的优点,使得算法同时兼顾了全局搜索和局部搜索的能力,根据个体优劣系数自适应调整交叉概率因子,改变以往交叉概率因子为定值常数,算法能够根据变异个体优劣选择合适的交叉概率因子.将改进算法用于以直流电机模型为被控对象的PID 控制器参数整定优化中.仿真实验研究表明,相较于传统PID,DE-PID 和QPSO-PID控制,AMDE-PID控制器具有更快的响应速度和稳定精度.%To solve the difficulty of PID controller parameters tuning in industrial processes,an adaptive mu-tation differential evolution algorithm was proposed according to analyzing the traditional differential evolu-tion algorithm and its application to the parameter optimization of PID controller. The algorithm defines pop-ulation similarity factor and individual quality factor, mutation operation is adaptive adjusted according to population similarity factor,which plays the advantages of different mutation operation mode and makes the algorithm both global search and local search ability,The crossover probability factor is adaptive adjusted ac-cording to individual quality factor,the algorithm changes the crossover probability factor is fixed value and chooses the appropriate crossover probability factor according to the variation of individual advantages and disadvantages. The algorithm was used to optimize PID controller parameters of the model of a DC motor. Experiments have shown that, Compared with the traditional PID control, DE-PID control and QPSO-PID control,the system has higher response speed and steady-state accuracy using AMDE-PID.