Font Size: a A A

Improved Particle Swarm Optimization Algorithm And Its Application Research

Posted on:2013-04-13Degree:MasterType:Thesis
Country:ChinaCandidate:Y Q DengFull Text:PDF
GTID:2248330374987613Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
As a new global optimization algorithm, Particle Swarm Optimization (PSO) which has characters such as simple concept and easy realization with only a few tuning parameters and so on, has been attracting more and more attention now and achieved primary success. Because of its short history, the theoretical research and application of PSO are still at initial stage. The paper gives a comprehensive study on PSO from the aspect of algorithm mechanism, modification and its application.Through analyzing the track of particle and problems in algorithm, we can know that PSO gets struck at local optima easily in its later iteration, which cause appearance of premature convergence. Therefore, this paper presents PSO based random disturbing(RDPSO), it increases population diversity and expands the search space through disturbing global optima, and then strengthens global detectability. However, this strategy still exits limitation that the ability of local search is still week. This algorithm is proved to be a overall convergent through analysis of convergence.To solve the problem of balancing good search area and high-grade global optimal solution with one strategy, a new algorithm of TS-RDPSO is proposed in this paper based on the openness of PSO and fusion of random disturbing and Tabu Search. The improved algorithm can increase population diversity through disturbing global optimal position, and then avoid effectively the mistake of sinking into local optima of PSO. At the same time, the proposed TS algorithm ensures its strong capacity of local search. The simulation results show that the improved algorithm is much better than PSO. In addition, the influence of the introduced parameters on the algorithm performance is analyzed and optimum is given.The key of Case-based Reasoning(CBR) is case retrieval, where Merits and Demerits of feature attribute weight affect directly the resulting. As a result, this paper utilizes optimal performance of the improved algorithm to optimize feature attribute weight and increase accordingly the accuracy of case retrieval. CBR is applied to concentration of copper ion prediction in arsenic-activated process, and the result show that, because of better optimization performance of improved algorithm, it can get accurate result and meet the application demand of practical engineering.
Keywords/Search Tags:Particle Swarm Optimization, Random disturbance, Tabu Search, Local Optima
PDF Full Text Request
Related items