Abstract: An improved Differential Evolution (DE) algorithm was proposed to solve the problem of premature convergence and improve the computational efficiency of DE. Firstly, different strategies with different parameter values were adopted to enrich the population diversity. Secondly, a new evaluation index was established to determine the suitable combination to match different phases of the search process. Finally, the evolution process was divided into many subprocesses to eliminate the negative effect of the previously selected combination. The contrast experimental results on ten classical Benchmark functions show that the proposed algorithm has a relatively better performance.

Key words: global optimization, differential evolution, evolution strategy, control parameter

摘要： 针对差分进化（DE）算法收敛早熟与计算效率不理想的问题，提出一种改进的差分进化算法。首先，在进化中同时并行多个策略与参数组合来提高个体多样性。其次，依据建立的评价指标自适应地调整组合来提高寻优效率。最后，把进化过程分为若干的子进程以避免前期优势组合不适应后期的问题。在10个标准测试函数上的实验结果表明，提出的算法与其他算法相比具有相对较好的性能。

关键词: 全局优化, 差分进化, 进化策略, 控制参数