The Study For Improvement Of Particle Swarm Optimization Algorithm

Particle Swarm Optimization Algorithm (PSO) is an Algorithm that mimicsthebehavior of flying birds and their means of information exchange to solveoptimizationproblems. The conception is easy and is easily realized, so it is accepted by internationalevolutionary computation researcher soon and applied in many field. PSO has become thefocus of research and hot issuesin the field of soft computing. However, particle swarmoptimization algorithmhas some shortcomings, such as slow convergence speed andsinking into local optima easily, to overcome these shortcomings, many researchers carrythrough analysis deeply from the perspective ofimprovinginertia weight and learningfactors or combining other algorithm, and they gain remarkable result.Based on previousstudies, this paper conducts deeply study in the improvement of particle swarmoptimization.At first, to control particles to fly inside search space and deal with the problems ofslow convergence speed and premature convergence of article swarm optimizationalgorithm, through the theoryanalysis and experiment proof about controlling particles tofly inside search space,this paper investigates boundary condition existing currently. Basedon the thought of round, this paper creatively proposes Boundary SutureParticle SwarmOptimization Algorithm. This strategy is stitching the largest and smallest boundaries inthe same dimension of the solution space of interest. It is like what two endpoints of a lineare coincided up and formed a circle. So the boundary is a closed area. This method issolving the particles flying out the search space. Experimental studies show BoundarySutureParticle Swarm Optimization Algorithm ensures the variety of particles and iseffective.Secondly, to overcome the shortcoming that the standard Particle SwarmOptimization (PSO) only can find a minimum of the objective function, we propose animproved PSO which not only can find all the minimums of the objective function, butalso has a strong global convergence. In each generation,through searching the extremeparticles in the population, we can get the extreme-point information of the objective function and achieve the purpose of searching more than one minimum of the objectivefunction. In the standard Particle Swarm Optimization, when searching the minimum ofthe multiple-minimum function, the global best particle may vibrate between differentminimums, our algorithm conquers this defect and improves the convergence and accuracy.The tests of some typical one-dimensional, two-dimensional and multi-dimensionalobjective functions show the improved PSO is effective.At last, this paper introduces the applications of particle swarm optimizationalgorithm in solving multiple equations. Boundary suture particle swarm optimization iscompared to the Standard Particle Swarm Optimization Algorithm.After solving threemultiple equations groups, the results identify Boundary suture particle swarmoptimization has a more strong overall optimization capability, numerical results illustratethe improved PSO has a strong superiority in solving multiple equations.