Firefly algorithm has shown good performance for solving various optimization problems. Most of the nature-inspired algorithms have the same problem which can easily trap in local optimal. To overcome this defect, a novel personalized step strategy for firefly algorithm (PSSFA) is presented. It uses large step for the optimal firefly and linearly decreasing step for the other fireflies to improve the ability of exploration. Experiments on 20 test functions show that the proposed algorithm can promote accuracy of the original method. Finally, we integrate PSSFA with k-means clustering for five datasets. The results show that PSSFA is an effective optimization algorithm.

Firefly算法在解决各种优化问题方面显示出良好的性能。大多数受自然启发的算法都存在相同的问题，很容易陷入局部最优。为了克服这一缺陷，提出了一种新颖的萤火虫算法个性化步调策略（PSSFA）。它为最佳萤火虫使用较大的步长，而对其他萤火虫使用线性递减的步长，以提高探测能力。对20个测试函数的实验表明，该算法可以提高原始方法的准确性。最后，我们将PSSFA与k-means聚类集成到五个数据集中。结果表明，PSSFA是一种有效的优化算法。