基于共轭梯度法改进的人工鱼群算法*

在基本人工鱼群算法中引入共轭梯度法,得到改进的人工鱼群算法。基本人工鱼群算法对每条人工鱼分别进行聚群算子和追尾算子,若更新结果没有得到改善则进行觅食算子,由于觅食算子具有很大的随机性,这样不能保证每次更新都是有价值的,本文用共轭梯度法代替此时的觅食算子,也就是如果人工鱼利用聚群算子和追尾算子更新,如果没有得到改善,利用共轭梯度法对该条鱼进行更新。在人工鱼群更新过程中引入共轭梯度法,减少随机性,增强人工鱼个体的局部寻优能力,确保人工鱼每次更新都会得到改善,从而加快人工鱼群算法收敛速度。数值试验结果表明,所得改进人工鱼群算法具有更快的收敛速度,同时收敛精度也得到一定提升。

Hybrid AFSA algorithm with conjugate gradient methods

Based on the basic artificial fish swarm algorithm, a modified artificial fish swarm algorithm is proposed by coupling the conjugate gradient. In the basic artificial fish swarm algorithm, the clustering operator and the rear-end operator are performed separately for each artificial fish. If the updating result is not improved, the foraging operator is carried out. Since the foraging operator has a lot of randomness, it can not guarantee that every update is valuable. In this paper, the conjugate gradient method is used to replace the foraging operator. That is, when the artificial fish is updated by using the clustering operator and the rear-end operator, if the result is not improved, the fish is updated by the conjugate gradient method. The conjugate gradient method is introduced in the process of artificial fish of updating, enhance the local optimization ability of artificial fish and reduce the randomness to ensure that the artificial fish will be updated every time, thus accelerates the convergence rate of artificial fish swarm algorithm. The numerical experiment results show that the improved artificial fish swarm algorithm has higher convergence speed, and the convergence accuracy is also improved.