基于双变异策略差分进化算法的模糊PI参数整定

针对模糊PID控制器参数难以整定的问题,提出一种基于双变异策略协同工作的自适应差分进化算法DSDE。该算法采用随进化代数变化的权重因子,将经由DE/target-to-best/1和DE/rand/2两种变异策略生成的个体加权组合成一个新的变异个体,并采用Z型函数根据迭代次数自动调整变异因子,以适应于不同的进化阶段。将DSDE算法应用于二阶被控对象的模糊PI控制器(FPI)参数整定,MATLAB仿真结果表明,与传统的FPI、DE-FPI和采用自适应变异差分进化算法进行参数整定的AMDE-FPI相比,基于DSDE算法的模糊PI控制器具有更好的控制性能。

Parameter tuning for fuzzy PI controller based on differential evolution algorithm with double mutation strategies

To solve the difficulty of parameter tuning for fuzzy PID controller, this paper proposes a self-adaptive differential evolution algorithm with double mutation strategies which is named as DSDE. It employs a weight factor varying with the number of iterations, creates a new individual by weighted combination of two mutated individuals via DE/target-to-best/1 and DE/rand/2 mutation strategies respectively, and uses a Z-type function to automatically adjust the mutation factor to adapt to different stages of evolution. DSDE algorithm is applied to the parameter optimization of a fuzzy PI controller (FPI) for a second-order controlled system. MATLAB simulation results show that DSDE-based fuzzy PI controller (DSDE-FPI) has better control performance than traditional FPI, DE-FPI and AMDE-FPI tuned by adaptive mutation differential evolution algorithm.