The Improvement And Application Of Particle Swarm Optimization

Particle Swarm Optimization (PSO) is an intelligent optimization algorithm, which simulates the collaborative optimization of individual in the flock bird. As a typical representative of swarm intelligent algorithms, PSO has been proved to be an effective global optimization algorithm. For its simple principle, easy implementation and less parameters, PSO attracted many researchers’attention as soon as it was proposed. At present, the PSO has been widely applied in engineering optimization, image processing, data control and other fields. But the PSO is derived from the biological community; it needs further research because of less theoretical basis and imperfect application. This paper mainly analyzes, improves and applies the standard PSO and discrete PSO, the main research work as follows:1. To overcome the problem of premature convergence in conventional particle swarm optimization, a new chaotic particle swarm optimization (ACPSO) with adaptive inertia weight is presented. The main idea of ACPSO is which first generates initial population with chaotic sequence based on the randomness and ergodicity of chaotic map; then calculates the evolution factor parameter using the idea of fuzzy classification and adjusts inertia weight dynamically based on the evolutionary state of the population. Compared with other algorithms, the ACPSO algorithm not only has great advantage of convergence property, but also avoids the premature convergence problem effectively, and at the same time, it shows the feasibility and validity of the ACPSO algorithm.2. The improved algorithm was applied to achieve image enhancement on the basis of the ACPSO’s performance. The experimental results show that the algorithm has a better effect in image enhancement compared with other algorithms.3. In view of a single real or binary particle swarm optimization which is unable to solve some engineering problems with both real and integral parameters, this paper presents a kind of hybrid real-binary particle swarm optimization (HPSO) algorithm. After the analyzing of algorithm, it is used to identify the sine signal parameters, and the experiment shows that the algorithm can achieve ideal results.4. For the traditional methods of solving multi-objective, which first transform multi-objective into single objective and then resolve the solution using the methods of deciding single objective problems. But this method usually has some defects, this paper introduces non-dominated sorting technology in ACPSO to solve multi-objective optimization and achieve multi-objective knapsack problem.