Dynamic Topology Hybrid Force Particle Swarm Optimization Algorithm And Its Application

As an efficient swarm intelligence algorithm, particle swarm optimization(PSO) algorithm is applied to various complex problems successfully. However, the accelerated restructuring and upgrading of Chinese economy under the New Normal bring about higher requirements for energy conservation and emission reduction of the production mode which needs to be optimized by each workflow. Yet, some of these problems are often difficult to be solved by using PSO algorithm directly. Therefore, force rules, population topology and the applications of practical production are studied for PSO algorithm in this paper.First of all, to overcome the defects of premature convergence and easily falling into local optimum which are caused by single mechanism of force rule between the particles, the particle swarm optimization algorithm is studied from the mechanism of force rules, and a new hybrid force particle swarm optimization(HFPSO) algorithm is proposed. The searching capability of HFPSO algorithm is tested through contrast tests by applying standard test functions, and its effectiveness is verified.Secondly, the particle swarm optimization algorithm is studied from the aspect of population topology based on four typical network models, which are globally coupled network model, nearest-neighbor coupled network model, ER model and small-wold network model respectively. The effects of different static and dynamic topology on performance and population diversity of the algorithm are studied through tests, and the relationships between characteristics of population topology and performance of the algorithm is obtained, which provides a basis for further research of unidirectional dynamic population topology suitable for HFPSO algorithm.Moreover, in order to reflect the interaction relationship between particles more accurately and simulate the behavior mechanism of the biological individuals that are willing to interact with better ones, a fitness-driven edge-changing unidirectional dynamic topology(FEUDT) structure is presented from the perspective of fitness selection. The FEUDT topology is conbined with the proposed HFPSO algorithm by simultaneously evolving of both algorithm and structure, an algorithm called dynamic topology hybrid force particle swarm optimization(DTHFPSO) algorithm is proposed. Furthermore, the parameters of the topological structure are discussed and the search ability of the proposed algorithm is tested through contast experiments.Finally, the proposed DTHFPSO algorithm is adopted to solve practical issues including reliability optimization of hydraulic system and scheduling optimization of hydraulic manifold proocessing shop, the results shows that the goals of cost saving and production efficiency are achieved. Meanwhile, the applicability of the DTHFPSO algorithm is proved.