Abstract: To improve the realcoded searching ability of the BiogeographyBased Optimization (BBO) algorithm, this paper presented a hybrid algorithm BBO/DEs based on Differential Evolution (DE). In order to solve the prematurity of BBO, the algorithm incorporated the search performance of differential evolution and the utilization performance of BBO. And a Levy mutation strategy was introduced to enhance population diversity. Finally suitable trial vector was chosen by comparison. Some correlation algorithms were compared on highdimensional benchmark functions. The experimental results show that without losing the original ability the proposed algorithm can improve the searching ability, and it has higher efficiency in terms of search accuracy and speed.

Key words: BiogeographyBased Optimization (BBO), Differential Evolution (DE), real coding, trial vector, Levy distribution

摘要： 针对生物地理学优化算法在实数编码时搜索能力较弱的缺点，提出一种基于差分进化的混合优化算法(BBO/DEs)。通过将差分进化的搜索性与生物地理优化算法的利用性有机结合，以解决原算法在局部搜索时容易出现早熟的问题；并构造一种基于Levy分布的变异方式，确保种群在进化过程中保持多样性；最后通过实验比较，选取了合适的试验策略。利用高维标准测试函数对相关算法进行实验，结果表明该算法能够克服搜索能力不足的缺点，并继承了原算法的快速收敛性能，可以有效兼顾精度与速度的要求。

关键词: 生物地理学优化, 差分进化, 实数编码, 试验向量, Levy分布