A Study On Particle Swarm Optimizers In Dynamic Environment And Their Applications

Optimization problem is ancient and challenging, which widely exists in scientificresearch and engineering practice. However, the traditional optimization methods can’t dealwith some complex optimization problems in reality because they cost the large amount ofcalculation. And sometimes, only local optima are attained. Recently, some computationalintelligent algorithms as GA, PSO and so on are employed to optimization problems for theirsimplicity and easiness to implementation. Some researches indicate that computationalintelligent algorithms are popular and effective in solving some optimization problems.Particle Swarm Optimizer (PSO) is first proposed by James Kennedy and Russel Eberhartin 1995, which is a relatively new swarm intelligent optimizer and inspired by bird flockflying behavior. PSO emulates the swarm behavior of insects, animals herding, birds flocking,and fish schooling, in which collaborative search for food exhibits a potential computationalmodel. PSO has many advantages, such as simplicity in concept, easiness to implement, globalsearch, robustness and so on. PSO has become the research focus in the field of computationalintelligence.PSO has been successfully applied to solve the static optimization problems. However,many problems are dynamic as time and space changes in real life. Therefore, optimalalgorithms should timely response and fast track the changing optimal in dynamicenvironment. This is not only a new research area, but also a challenge for particle swarmoptimizer.This paper conducts research on dynamic PSO from its principle, method and application.The main contributions are as follows.(1) A new self-adaptive particle swarm optimizer (AVPSO) is proposed for overcomingPSO’s disadvantages, namely it often convergences in local optima. Firstly, the global optimalfitness value and global worst fitness value of individual particles is found out. Then，"social"influence of the standard particle swarm optimizer is replaced based on the above two valuesand at the same time，a new response for environment changes is designed and incorporated.At last, AVPSO is verified by experimental on many functions.(2) A Dynamic Particle Swarm Optimizer with Adaptive Diversity Preservation (GSPSO) is proposed to deal with how to keep population diversity in the evolutionary process. Thewhole population is divided into two subswarms. A subswarm based on PSO is devoted todevelop a new search area for optimal solution and the other based on GSO keeps populationdiversity so as to avoid to convergence to local optima. The experiment results show that theimproved algorithm has good performance to keep population diversity.(3) A group animation path planning subsystem based on improved dynamic PSOs wasdesigned and implemented. Collision detection and avoidance in the process of path planningare designed first. Improved dynamic PSOs are applied to implement phenomenon ofanimation path planning. The path data are used in an animation flash.