Research of Artificial Fish Swarm Algorithm with Propagatable Ability

By analyzing the disadvantages of AFSA, an improved algorithm called propagatable artificial fish-swarm algorithm (PAFSA) that adds the propagatable ability to some artificial fishes is presented. In the algorithm, the fish swarm is classified based on similarity so that the excellent individuals can carry out local optimal solutions' search and poor individuals can carry out search spaces' locating in order to find-new spaces of local optimal solutions, which can merge the properties of local and global search. The even cross operation, variation operation and moving weight adjustment are used to improve all kinds of behaviors of artificial fishes. The algorithm has the advantages of simplicity in calculation and easiness in implementation as AFSA, and can improve the ability of jumping out the local optimal solutions and gain a high solving precision. Through the BUMP problem's solving experiment, the algorithm has good effect.