Abstract：A hybrid particle swarm optimization(GMCMPSO)algorithm combining gravity measure and centroid mutation strategy is proposed to solve the problem that the classical particle swarm optimization(PSO)algorithm tends to fall into the local extremum in the early stage and the convergence accuracy is low in the late stage.Firstly，elite grouping strategy is adopted in the initial phase of the algorithm to obtain the excellent information of the population.Secondly，two subgroups are measured by gravity to achieve efficient information sharing between populations.Finally，some ordinary particles are randomly mutated under the guidance of gravity measure，and the rest of ordinary particles are mutated at the center of mass，so that the algorithm can effectively jump out of the local extreme value and develop the most potential region，and improve the convergence accuracy of the algorithm.The proposed algorithm and classical particle swarm optimization(PSO)algorithm and hybrid fireflies and particle swarm optimization(HFPSO)algorithm based on hierarchical autonomous learning，improved particle swarm optimization(HCPSO)algorithm，fitness depends on optimization(FDO)algorithm，a total of five in sixteen on the standard test functions are compared，the experimental results show that the GMCMPSO on high-dimensional multimodal function compared with other four algorithm has higher convergence accuracy and faster convergence speed.
2022, Vol. 43 
