Abstract: In order to overcome the slow convergence, low precision and local optimum of basic artificial fish swarm algorithm(AFSA), this paper proposes a novel artificial fish swarm algorithm. This algorithm takes advantage of individual learning ability and social learning ability of particles in artificial fish swarm and particle swarm, simulates the speed and position of particles in particle swarm to update and modify the formulas of foraging, grouping and tailing. It then experiments five typical testing functions, analyzes the optimization precision, convergence speed and stability of the algorithm. The results turn out that the improved algorithm has stronger convergence, more stability and better performance.

Key words: artificial fish swarm; particle swarm; intelligent optimization; individual learning; social learning