基于群体行为的自适应变异算子鱼群算法

The Self-Adaptive Mutation Artificial Fish-School Algorithm Based on Group Behaviors

针对人工鱼群算法的不足,考虑了包括鱼群个体之间的相互感知作用、群体的领导模式并结合萤火虫群中个体的光强吸引度在内的群体行为特点来对鱼群行为进行完善。同时,在算法改进方面,采用了自适应步长和视野,并且引入了Gauss变异算子和遗传算法在一定情况下对鱼群个体进行变异操作。在此基础上,提出了一种新型自适应变异算子的鱼群算法。通过典型函数验证结果表明该算法在收敛速度、精度、稳定性及克服早熟能力方面都有了显著的提高。

According to the disadvantages of the Artificial Fish-school Algorithm, the fish-school behav- iors are consummated by considering the group behaviors of neighborhood sensing factors, leader mode and the attraction of the individuals in the glowworm swarm. Meanwhile, it makes an improvement in combining the self-adaptive step and visual, Gaussian mutation and Genetic Algorithm together. Based on these improvements mentioned before, a new Artificial Fish-school Algorithm based on self-adaptive mu- tation operation is raised. By using typical functions to examine, the simulation results show that the con- vergence speed, optimization precision, algorithm stability and the ability to avoid precocious phenome- non of the improved algorithm are much better than the standard one.