Hill Valley Function Based Niching Particle Swarm Optimization for Multimodal Functions

A novel niching Particle Swarm Optimization (PSO) method based on a hill valley function is proposed. In this algorithm, the hill valley function is used to decide whether the niching seed particle and its neighbour are on the same hill, and if they are, a new niching is formed. The hill valley function is also used to decide whether two niching subswarms are on the same hill, and if they are, the two niching subswarms are merged. The proposed algorithm is evaluated using three benchmark test functions. Results indicate that the proposed hill valley function based niching PSO algorithm has strong adaptive searching capability and efficient convergence in searching multiple solutions.