Research On Improvement Of Particle Swarm Optimization Algorithm And Its Application In Solving Semi-infinite Programming Problems

Particle swarm optimization algorithm is an optimization algorithm based on swarm intelligent.It is simple in principle,few in parameters and easy in implementation.It has been widely used to solve practical and mathematical problems,including mathematical programming problems.Semi-infinite programming problem is a kind of non-smooth mathematical programming problem,which has important applications in many fields.This thesis gives a brief overview of particle swarm optimization and semi-infinite programming problem.On this basis,researches on improvement are carried out aiming at the shortcomings of particle swarm optimization algorithm.All particles are divided into different subgroups according to their evolution abilities,and different evolution strategies are executed according to the characteristics of different subgroups to increase the diversity of the population.A multi-strategy particle swarm optimization algorithm based on evolution ability is established.At the same time,the inertia weight with random fluctuations is designed in the algorithm,so that particles still have the ability to jump out of the current area in the later stage of the algorithm.Numerical experiments are carried out on the test functions,and the results show that the algorithm has improved both the convergence speed and the solution accuracy.In the particle swarm optimization algorithm,the second and third best individuals in the population and the randomly selected excellent individuals are introduced,and a variety of learning methods are generated through the adjustment of the proportional coefficient,thereby increasing the learning objects and learning strategies.Combining it with the grey wolf optimization algorithm that introduces randomly selected excellent individuals,and introducing a multi-round perturbation mechanism into the algorithm,a multi-learning object hybrid PSO-GWO algorithm is established.Numerical results show that the proposed algorithm has better performance than the comparative algorithms.The use of particle swarm optimization algorithm to solve the semi-infinite programming problem is studied.Two models of exact penalty function and maximum entropy function are used to transform the semi-infinite programming problem,and the multi-strategy particle swarm optimization algorithm based on evolution ability,the multi-learning object hybrid PSO-GWO algorithm and related algorithms are used to solve the transformed problem.Numerical experiments are carried out on several semi-infinite programming problems,and the results show that both methods are feasible and effective.This thesis has 5 figures,11 tables and 101 references.