Research On Multi-objective Particle Swarm Optimization Algorithm Based On Grid Strateg

The multi-objective optimization problem has the characteristics of complexity and multi-polarization,and the classical algorithm can not solve it effectively.However,with the development of information technology,swarm intelligence algorithm emerges as The Times require.This kind of algorithm provides new ideas for solving multi-dimensional and multi-objective complex problems.As a kind of swarm intelligence algorithm,particle swarm optimization has the characteristics of fast convergence,simple operation,easy implementation and high computational efficiency,so it has been successfully applied in many multi-objective optimization problems.When particle swarm algorithm deals with multi-objective optimization problems,it is called multi-objective particle swarm algorithm.However,because of the fast convergence speed of the multi-objective particle swarm optimization algorithm,it is easy to make the population gather and fall into the local optimal solution.In order to improve the diversity and convergence of the algorithm,the multi-objective particle swarm optimization algorithm is improved.The main work is as follows.(1)In order to enhance the diversity of particles in the multi-objective particle swarm algorithm during the optimization search,The multi-objective particle swarm algorithm based on the adaptive grid mixing mechanism(ammm MOPSO)is proposed.The algorithm uses a dual maintenance strategy of adaptive grid and hybrid mechanism to ensure the non-inferior solutions in the external archive are uniformly distributed to avoid rapid degradation of the population,which affects the particle exploitation capability.Using the weighting strategy in the hybrid mechanism to determine the global optimal sample among the non-inferior solutions archived externally increases the diversity of the population and enhances the probability of particles flying to the true Pareto front.Also,to prevent the algorithm from stagnating and falling into the problem of local optimum,a variation operation is introduced to dynamic variation of the particle positions,which enhances the exploration ability of the particles.The simulation experimental results show that the proposed algorithm has better convergence and diversity compared with other three international classical multi-objective particle swarm algorithms,and better spatialization effect.(2)In order to improve the convergence of multi-objective particle swarm algorithm particles,the interval of the number of non-inferior solutions with relatively good algorithm effect is obtained by experimental simulation.Multi-objective particle swarm optimization algorithm based on double random(DRMOPSO)is proposed.The algorithm controls the number of non-inferior solutions in a better interval by using a grid technique and a dual randomization strategy with mixed ranking to make the algorithm get better convergence and diversity.In order to prevent population degradation,which affects the particle exploitation ability,the proposed grid congestion distance strategy makes the non-inferior solutions in the external archive uniformly distributed to improve the probability of particles flying to the true Pareto front.The simulation results show that the proposed algorithm has better convergence and diversity compared with MOPSO.(3)In order to improve the convergence and diversity of MOPSO,the proposed The grid-based technique and multi-strategy multi-objective particle swarm optimization(GTMSMOPSO)are proposed.The algorithm randomly uses one of two different evaluation metric strategies(convergence evaluation metric and distribution evaluation metric)combined with the grid technique to enhance the diversity and convergence of the population and improve the probability of particles flying to the true Pareto front.Adopting a grid technique is used to maintain the external archive in combination with a hybrid evaluation metric strategy to avoid removing particles with good convergence based only on particle density,which leads to population degradation and affects particle exploitation capabilities.The simulation experimental results show that the proposed algorithm has better convergence and diversity than CMOPSO,NSGA-II,MOEA/D,MOPSOCD and NMPSO.The proposed algorithm has better convergence and versatility than CMOPSO,NSGA-Ⅱ,MOEA/D,MOPSOCD and NMPSO.