针对差分进化算法的优化性能受变异策略的影响,提出了一种组合差分进化算法,提高了算法的优化性能. 该算法将差分进化算法常用的四种变异策略分为携带最优值信息和不携带最优值信息两类. 在进化过程中,从这两类中随机各选取一种策略参与进化. 通过五个测试函数,证实了该组合算法的有效性. 在此基础上,将该组合差分进化算法对PID控制器参数进行整定优化. 通过仿真实验研究,表明采用组合差分进化算法获得的PID控制器性能优于基本差分进化算法设计的PID控制器.%The performance of conventional differential evolution( DE) algorithm depends on mutation strat-egy. An ensemble differential evolution ( EDE ) algorithm is proposed. The performance of EDE is im-proved. Four mutation strategies are divided to two groups. The first group is with best information and the second group is without best information. The two mutation strategies are randomly selected from two groups respectively. The validity of the proposed algorithm is proved by five testing functions. An optimization method of PID controller parameters based on EDE is provided. Experiments have shown that the perform-ance of the PID controller of EDE is superior to that of the conventional DE.
展开▼
