Research On Multi-objective Problem Based On Particle Swarm Optimization

In recent years,the use of particle swarm optimization(PSO)to solve multi-objective optimization problems has become a hot topic in academic research.Compared with evolutionary algorithm,PSO has many advantages,such as simple principle,fewer parameters,and no need to use crossover and mutation operations.However,there are also some defects in the use of multi-objective PSO.In the face of many problems,PSO has many advantages.In the case of local optimal location,the performance of multi-objective particle swarm optimization algorithm is not satisfactory.Therefore,it is urgent for researchers to find more efficient algorithms.In order to further improve the diversity and convergence of the algorithm,and improve the ability of the algorithm to jump out of local optimum,this paper proposes an improved multi-objective particle swarm optimization algorithm based on the existing principles of multi-objective particle swarm optimization.Firstly,the multi-objective optimization problem is introduced in detail,and the basic concepts,design objectives and evaluation indexes of the multi-objective problem are introduced in detail.Then the basic particle swarm optimization(PSO)algorithm is introduced and compared with the classical multi-objective particle swarm optimization(MPSO).The reason why the algorithm is easy to fall into local optimum is analyzed,and the competitive strategy based PSO algorithm is introduced.This paper introduces the linear decreasing inertia weight method which is widely used at present,and points out its advantages and disadvantages,that is,it can adjust the global multi-line of the algorithm and propose an improved diversity detection strategy to quantitatively respond to the diversity change of the population,and dynamically change the inertia weight value in the particle update formula by using the diversity rating index of the population,so that the change of the inertia weight of the particle is not only the change of the inertia weight of the particle.In the form of random selection,the current state adaptability of the population is changed.At the same time,the population decomposition strategy is proposed.The idea of population decomposition strategy is inspired by the traditional linear decreasing inertia weight method.In the early stage of population iteration,the inertia weight of particles is larger in order to obtain more diversity of algorithms.In the later stage of algorithm iteration,smaller inertia weight values are needed in order to obtain stronger convergence.Enhance the accuracy of particle optimization.Therefore,the population decomposition strategy is combined with the dynamic inertia weight strategy based on diversity detection.The population is decomposed into two populations,one named as diversity population,which uses random inertia weight to enhance the diversity of particles,and the other named as convergence population.The inertia weight is adjusted by population diversity evaluation index to enhance the algorithm.Astringency.This paper adopts a series of test functions to test the performance of the algorithm.By comparing with the improved multi-objective particle swarm optimization algorithm and the traditional multi-objective evolutionary algorithm in several diversity evaluation indicators,the overall evaluation of the performance of the algorithm is obtained,and good results are obtained.It further verifies the diversity and convergence of the improved algorithm and jumps out of the local optimal position.It has good performance in these capabilities.