An Improved Parameter Control Based on a Fuzzy System for Gravitational Search Algorithm

Recently, a kind of heuristic optimization algorithm named gravitational search algorithm (GSA) has been rapidly developed. In GSA, there are two main parameters that control the search process, namely, the number of applied agents (Kbest) and the gravity constant (G). To balance exploration and exploitation, a fuzzy system containing twelve fuzzy rules is proposed to intelligently control the parameter setting of the GSA. The proposed method can enhance the convergence ability and yield better optimization results. The performance of fuzzy GSA (FGSA) is examined by fifteen benchmark functions. Extensive experimental results are tested and compared with those of the original GSA, CGSA, CLPSO, NFGSA, PSGSA and EKRGSA.