Adaptive particle swarm optimization algorithm based on dynamic link matrix and its application

To deal with the problems of topological structure cannot adjust adaptively, easy to trap into the local minimum and diversity losing in traditional particle swarm optimization algorithm, a newly adaptive PSO algorithm based on dynamic link matrix was proposed, which build the neighborhoods though link matrix and divide them into the sub-swarm based on feature clustering. The algorithm can adjust the link probability between particles according to the evolution states of different sub-swarms, which not only realize the topology structure adjustment adaptively and every sub-swarm evolves as their own evolutionary state, but also keep the swarm diversity and better character. The simulation results of the standard benchmark test functions show that the proposed algorithm is better and more effective than the topology of the other proposed PSO algorithm. The feasibility of the method is illustrated with the challenge of the optimization of the selectivity of the first reaction in Kumar.