Composite Particle Swarm Algorithm And Its Application

Abstract:Global optimization problems exit in many fields such as science, engineering and commerce, it can be described as a D-dimensional minimization problem simply. But unfortunately, many of these problems can’t be solved by traditional analytical methods due to their complexity in real life, so numerical algorithms such as evolutionary computation algorithms are generated and adopted to slove them. But the biggest challenge is that these algorithm are easy to be trapped in the local optima of an object function, especially in high dimensional and complex multimodal problems. Particle swarm optimization(PSO) is one of these numerical algorithms to slove global optimization problems. PSO also has been shown to perform well on a lot of optimization problems. But PSO suffers from the shortcoming that it may easily get trapped in a local optimum because of the early loss of population diversity when sloving complex problems. Many attempts has been adopted to improve the performance of the original PSO, and some state-of-the-art original PSO variants are proved effective on complex problems but they usually have a relatively low convergence speed.In order to improve the performace of the original PSO both on the solution quality and convergence speed, this thesis presents a composite particle swarm optimization(CPSO) using a "novel learning strategy plus one assisted search mechanism" architecture. The proposed learning strategy called combination leaning strategy, this stratgy combines one particle’s historical best information and the global best information into one learning exemplar to guide the particle movement, it can reserve the original search information and lead to faster convergence speed. The proposed assisted search mechanism is designed to look for the global optimum. Search direction of particles can be greatly changed by this mechanism so that the algorithm has a large chance to escape from local optima. In order to make the assisted search mechanism more efficient and the algorithm more reliable, the executive probability of the assisted search mechanism is adjusted by the feedback of the improvement degree of optimal value after each iteration.Then the composite particle swarm optimization algorithm is applied to function optimization problems and vehicle routing problem. In function optimization problem,22benchmark functions were used to test the performance of the algorithm. In vehicle routing problem, the VRP and VRPTW problems were used to test the performance of the algorithm. Through these simulations, the high efficiency of composite particle swarm algorithm is verified.