Abstract: After analyzing the low local search ability of Artificial Fish Swarm Algorithm (AFSA), an improved AFSA based on simplex method was proposed. In the latter evolution period, the improved algorithm used simplex operators that were distributed evenly and widely as its artificial fishes to replace the original fishes which got together around the local optimum solution in every some generations. By adding simplex operators to AFSA, the level of detailed search was greatly improved in local part. Finally, the improved algorithm is proved to be a more effective algorithm in solving the problem of the low optimization accuracy by using three typical test functions.

Key words: Artificial Fish Swarm Algorithm (AFSA), simplex method, simplex method operator, optimization, accuracy

摘要： 针对鱼群算法在局域搜索能力差的问题,提出一种基于单纯形法的改进型人工鱼群算法。利用单纯形算子在局部区域内分布更均匀且广泛的特征,在鱼群算法运行到后期时,将单纯形算子每隔一定代数引入到现有的鱼群算法中取代原来大量聚集在非极值点附近的人工鱼,有效改善个体质量,提高局部搜索精细度,进而提高算法的寻优精度。采用典型算例对算法性能进行了验证分析,研究结果表明,该算法在解决鱼群算法后期优化精度低问题时可以获得更好的效果。

关键词: 人工鱼群算法, 单纯形法, 单纯形算子, 优化, 精度