Research On Multi-objective Optimization Problem Based On Improved Particle Swarm Optimization

Multi-objective optimization problems not only have practical and extensive applications in many fields such as engineering practice and aviation design,but are also ubiquitous in life.For example,the layout of the city where people live,various logistics and distribution,the behavior paths of Internet product user feedback.It occupies a prominent position in the field of optimization.The various objectives in multi-objective optimization problems usually restrict and conflict with each other,which means that it is impossible for all objectives to achieve the optimal situation at the same time.The above effectively solves multi-objective optimization problems brings certain difficulties.In recent years,intelligent algorithms have made great progress in the study of multi-objective optimization problems.However,when faced with some complex problems,the results are usually uneven,and the obtained solutions often converge prematurely or are unevenly distributed.In this dissertation,the multi-objective particle swarm optimization is effectively improved to improve its convergence.The main contents include:(1)A novel multi-objective particle swarm optimization with cosine distance strategy and game strategy are proposed.The approximate solution of the Pareto set in external archive is updated using the cosine distance.At the same time,a candidate set is established to effectively replace the non-dominated solutions deleted from the external archive,which is beneficial to the maintenance of external archive and improves the convergence of the solution and the diversity of the population.In order to enhance the selection pressure of the global leader,a method combining cosine distance and game strategy to select the global leader is proposed.In addition,mutation is used to maintain the diversity of the population and prevent the population from prematurely converging to the true Pareto front.(2)A multi-objective particle swarm optimization with dynamic population size is proposed.This paper uses the multi-swarm approach to solve the problem of premature convergence of the particle swarm optimization and promote the diversity of the population.It adopts the adaptive strategy of population growth to increase the population size,and uses the reverse search strategy to increase the diversity of the population.Then,the distribution estimation and evaluation strategy of population reduction is adopted.On the one hand,the distribution of all non-dominated solutions is evaluated,and the global leader is selected;On the other hand,the non-dominated solutions with poor distribution are removed,the population size is reduced,and the population size is prevented from excessively increasing.By dynamically adjusting the size of the population,it can provide the computational resource requirements of the algorithm at different stages.At the same time,it promotes the competition between populations and converges to the global optimum,thereby maintaining the diversity of populations.