A new differential evolution algorithm (DE) is proposed in this paper to overcome the problem of premature convergence and low search efficiency of the original DE. In the new algorithm, several mutant individuals are generated simultaneously in each population using different strategies with some specified parameters, which effectively maintains population diversity and avoids premature phenomenon in the evolution. According to the survival of the fittest law, the trial individual with the best fitness value will enter the selection stage to generate the new population, which improves the search efficiency. The contrast experimental results with DE and its improved algorithm show that the effectiveness of the proposed algorithm. The algorithm is also applied into the fuzzy cluster analysis and solves the problem of falling into the local extreme of the original cluster model.%针对差分进化算法早熟与搜索效率不理想的问题,提出一种改进的差分进化算法.算法在变异阶段采用多策略与多参数并行的方法一次产生多个变异个体,有效地保持了种群中个体的多样性,抑制了早熟现象的发生.根据竞争机制选择适应度最好的变异个体进行选择操作,提高了搜索效率.与差分进化及其改进算法的对比实验表明了算法的有效性,并把提出的算法应用到模糊聚类分析中,较好的解决了原始聚类模型求解容易陷入局部极值的问题.
展开▼