An effective approach for adaptive operator selection and comparison for PSO algorithm

Abstract The search process with metaheuristic algorithms is mostly performed using one operator. The most important problem of using only one operator in the algorithm is that the success of the algorithm is determined by the success of the operator used. If the selected operator fails, it can be said that it is very difficult for the algorithm to be successful. To improve the algorithm's performance, the number of operators can also be increased. Using a total of three operators, a particle swarm optimization technique is suggested in this paper to solve 28 problems, comprising 5 Unimodal functions, 15 Multimodal functions, and 8 Composition functions in the CEC 2013 benchmark problems. In the proposed algorithm, parameter tuning operations were performed to determine the optimal parameters. Then, Adaptive Pursuit and Probability Matching methods were used to select the most successful operator with the optimal parameters. The obtained data were compared with eight different algorithms in the literature. It was observed that the proposed algorithm was more successful than the compared algorithms in 30 and 50 dimensions and showed a competitive behavior in 100 dimensions.