Abstract: A new adaptive Glowworm Swarm Optimization algorithm with Changing Step（CSGSO） is presented to solve the problem that the Glowworm Swarm Optimization（GSO） algorithm to optimize the multi-modal function existing slow convergence and low precision defects. The successful or failure to search is introduced in this new algorithm. In each iteration process, the step is changed dynamically according to the searching is successful or failure, which provides the algorithm with effective dynamic adaptability. Experimental results show that, the CSGSO can effectively improve the GSO algorithm to optimize the multi-modal function existing slow convergence and low precision problems. Compared with other algorithms, the CSGSO algorithm has the advantages of simple operation, easy to understand, fast convergence rates and high precision and so on.

Key words: multimodal function optimizing, Glowworm Swarm Optimization（GSO）, adaptability, Glowworm Swarm Optimization algorithm with Changing Step（CSGSO）, multimodal function

摘要： 针对萤火虫群优化（GSO）算法优化多模态函数存在收敛速度慢和求解精度不高等缺陷，提出一种变步长自适应萤火虫群优化算法（CSGSO）。该算法主要思想是在GSO算法中引入搜索成功与失败概念，在每次迭代中萤火虫个体据其搜索成功或失败，加大或减小其搜索步长，使算法具有动态自适应性。实验结果表明，该算法可有效地解决GSO算法优化多模态函数存在收敛速度慢和求解精度不高的问题，增强了GSO算法优化多模态函数的性能；与其他算法相比，提出的算法具有操作简单、容易理解、收敛速度快和求解精度高等优点。

关键词: 多模态函数优化, 萤火虫群优化（GSO）, 自适应, 变步长萤火虫群优化（CSGSO）, 多峰函数