A Knowledge-Based Artificial Fish-Swarm Algorithm

The Artificial Fish-swarm Algorithm (AFA) is an intelligent population-based optimization algorithm inspired by the behaviors of fish swarm. Unfortunately, it sometimes fails to maintain an appropriate balance between exploration and exploitation, and has a drawback of blind search. In this paper, a novel cultured AFA with the crossover operator, namely CAFAC, is proposed to enhance its optimization performance. The crossover operator utilized is to promote the diversification of the artificial fish and make them inherit their parents´ characteristics. The Culture Algorithms (CA) is also combined with the AFA so that the blind search can be combated with. A total of 10 high-dimension and multi-peak functions are employed to investigate the optimization property of our CAFAC. Numerical simulation results demonstrate that the proposed CAFAC can indeed outperform the original AFA.