Multi-objective Particle Swarm Optimization Algorithm Based On External Population Self-adaptation And Its Application

In recent years,people have been widely used in real life and the impact of actual production problems.The multi-objective optimization problem has received increasing attention in the engineering and academic circles,and many methods for dealing with problems have been proposed.At present,the research of intelligent optimization algorithms is the most popular,and has been applied to people's production and life.Inspired by bird predation or predation behavior,the scholar proposed that Particle Swarm Optimization Algorithm(PSO)is a typical swarm intelligence optimization algorithm.The PSO algorithm is simple in structure,easy to implement,and fast in convergence,which has attracted the attention of researchers.It is widely used to solve optimization problems as well as practical production management,power control and many other issues.However,particle swarm optimization algorithms still have many problems in solving multi-objective optimization problems such as low convergence accuracy and poor solution diversity.In order to overcome these problems,this paper proposes three improvement strategies and is used to solve multi-objective optimization problems.The work involved in this paper mainly includes:1.The basic particle swarm optimization algorithm may converge to the local optimal Pareto frontier.An adaptive strategy is proposed to solve the problem of low convergence precision and uneven distribution of multi-objective particle swarm optimization algorithm.This strategy can improve the convergence speed and convergence precision of the algorithm.It can be seen that modifying the three control parameters(such as inertia weight and two learning factors)can directly affect the convergence performance of the particle swarm optimization algorithm.Therefore,a reasonable setting of these three parameters can improve the performance of the particle swarm algorithm.The three control parameters(inertia weight and two learning factors)of the particle swarm algorithm are adaptively set to balance the local search and the global search.2.Although the decomposition-based update strategy can maintain the diversity of the population well,when dealing with the problem that the Pareto frontier surface is degraded or cannot completely cover the unit hyperplane,the diversity of the solution cannot be well maintained.To overcome this problem,we have designed an improved update strategy to maintain the quality of the external population.3.A selection strategy is designed to determine the global optimal position(gbest)and local optimal position(pbest)of each particle to help the particle swarm operator to perform global and local searches.Choosing the performance of strategy local search and global search has a very important impact.Therefore,an appropriate selection strategy can improve the performance of the algorithm.For different purposes,the selection strategies are different from each other.In this paper,the selection strategy is used to select the solution generation children.The main goal of the selection strategy is to help the crossover operator implement global search and local search.The SMPSOE algorithm is applied to the anti-collision problem of the car and compared with other algorithms.The experimental results show that it is superior to other algorithms.The experimental verification by classical test function and comparison with MPSOD and dMOPSO,MOEA/D,MOEA-DVA,the SMPSOE algorithm has improved in terms of convergence,diversity and distribution.