改进多种群差分进化算法的混沌系统参数估计

摘要

针对混沌系统参数估计的多峰寻优问题，提出一种改进的多种群差分进化算法。改进差分进化算法的变异操作，使其前期更适合全局性搜索，利用α核心集对当前种群进行聚类，分别对聚类后的子群选用贪婪的差分变异算子完成深度搜索，比较所选取各子群的最优值，得到全局最优值作为是否结束搜索的判断依据，并将其应用到混沌系统参数估计中。实验结果表明，该算法对于多峰值、大空间的全局性参数估计在收敛速度、精度上优于混合量子进化算法、改进粒子群优化算法以及DE/best/2算法。%In order to solve the multimodal optimization problem in chaotic systems parameter estimation,an improved multi-swarm Differential Evolution( DE) algorithm is proposed. The mutation operator of DE algorithm is improved, which is more suitable for the global search. By usingαcore set clustering the current swarm,the depth search with greed DE operator is completed on clustered swarms respectively. By comparing the optimal values of selected swarms, the global optimal value is obtained as the judgment of whether to end the search,and is applied to the parameter estimation of chaotic systems. Experimental results show that the proposed algorithm is better than the Hybrid Quantum Evolutionary Algorithm(HQEA),Improved Particle Swarm Optimization(IPSO) and original DE/best/2 algorithm in convergence rate and accuracy for multi peak,large space of global parameter estimation.