Abstract: A novel differential evolution algorithm is proposed to improve the exploration and exploitation capabilities of differential evolution（DE）. In this algorithm, a new mutation operator following the t distribution is used to integrate the advantages of Gaussian and Cauchy mutation, while both the mutation strategy and crossover probability can be gradually self-adapted by learning from their previous successful experience. Experimental studies are carried out on four classical Benchmark functions, and the computational results show that the algorithm has fast convergence, high accuracy more robustness.

Key words: differential evolution, t distribution, mutation strategy adaptation, crossover probability adaptation

摘要： 为了更好地提高差分进化算法的全局探索和局部开发能力，提出了一种改进的差分进化算法。在该算法中，引入t分布变异算子将高斯变异和柯西变异的优点结合起来，根据以往的进化经验自适应地调整进化策略及交叉概率。通过四个典型的Benchmarks函数的测试结果表明算法具有良好的性能。

关键词: 差分进化, t分布, 进化策略自适应, 交叉概率自适应