人工鱼群算法的改进

Improvement of Artificial Fish Swarm Algorithm

人工鱼群算法(AFSA)是一种新型随机搜索优化算法。通过初步的研究表明,该算法具有许多优良的性质,但也有一些不足之处。针对均匀随机行为和常数拥挤度因子而导致算法运行时间长或陷入局部最优的问题,引入对称正态随机行为,自适应调整该行为参数,减少了由于迂回搜索导致的无用计算,也使人工鱼可在解空间进行更为广泛的搜索,提高了搜索效率。采用了自适应拥挤度因子并提出新的适应度函数,加快了系统满意解的收敛速度,使数值解更加稳定。实验结果表明,与基本人工鱼群算法相比,该方法具有明显的优越性。

Artificial Fish Swarm Algorithm （AFSA） is a new random search optimization algorithm. The preliminary study shows that it has many promising features, but also some disadvantages. Aiming at the problem of AFSA, such as long running time or being in local optimal, caused by uniformly random behavior and constant of congestion factor. Based on symmetric normality random behavior, self-adaption adjusts the parameter of this behavior, and a large number of unused circuitous searches are reduced, and a more complete search within solution space is obtained for artificial fishes so that a high search efficiency is arrived at. The self-adaption congestion factor is adopted and a new fitness function is porposed,increasing the convergence rate of satisfactory solution domain,making the result more stable. Results of experiments show that there is an obvious advantage for this improved method compared with the basic artificial fishswarm algorithm.