Research On Particle Swarm Optimization Algorithm With Enhanced Searching Ability

In the real world,there are many optimization problems.The optimization problem can be defined as finding a set of solutions to meet the requirements of a system for some performance indicators under some constraints.With the development of society,people need some tools to solve these optimization problems,which resulting in a variety of optimization algorithms.The particle swarm optimization(PSO)algorithm is a new optimization technique,which is based on the foraging behavior of birds.It is widely used in solving optimization problems because of its simplicity and practicality.When solving the optimization problem with the PSO,it is unnecessary for people to know the characteristics of the problem.The PSO can find the optimal value by updating a set of feasible solutions generation after generation by using of some strategies.With the characteristics of simple and high efficiency,the PSO algorithm has attracted attention of scholars both at home and abroad.But in the process of research,it is found that the PSO algorithm has some defects like premature convergence and low precision of search.In an attempt to eliminate the defects of PSO algorithm,this thesis proposes two schemes for improvement:1)In the first scheme,the particles are firstly divided into two groups,which are used for global search and local search respectively.This is similar to the multi-population PSO algorithm,but the difference is that the number of particles in these two groups can change dynamically.Under the premise that the total number of particles is constant,most of the particles are used for global search in the early stage,and as the iteration proceeds,the particles will slowly turn to the local search.In order to enhance the global searching ability,a new strategy will be used in this scheme to update the position of the particles in global search.In order to enhance the local search ability,the strategy of small scale variation will be adopted for particles engaged in local search.And those two groups of particles are not isolated from each other,unique information exchanges may occur between them.2)The second scheme is put forward under the inspiration of a common social phenomenon.In this scheme,with the core idea of “letting the best particle become more excellent and the worst particle try to be excellent.” The directional mutation strategy is adopted for the best particles and the worst particles in the population so as to improve the performance of the PSO algorithm.The importance of introducing the worst particle variation can be verified through the comparison experiments.Test functions show that these two schemes can effectively enhance the searching ability of the particles and alleviate the premature convergence.Finally,satisfactory experimental results are obtained after applying the particle swarm optimization algorithm with directional mutation to PID parameter adjustment.