Research And Application Of Multi-objective Particle Swarm Optimization Algorithm

In industrial production field and real life, we always encounter with multi-objective optimization problems. There are tones of differences between multi-objective optimization problem and single objective optimization problem, the most significant one being that the solutions of multi-objective optimization problem are not a single one but a solution collection in which all solutions are relatively eclectic and non-dominated solutions. There exist various defects in classical methods to solve multi-objective optimization problem, i.e., in classical methods, it is relatively difficult to find global optimal solution and diversity of solutions is not ideal enough, etc. There are plenty of advantages in Particle Swarm Optimization algorithm, which has become a new research hot spot in intelligent optimization ever since launched. Current research results reveal that Particle Swarm Optimization algorithm has been widely used in various fields such as optimization of Multi-objective function.This thesis introduces method and algorithm of Particle Swarm Optimization algorithm, analyzes research conditions in solving multi-objective optimization problems with particle swarm optimization algorithm and makes a deep exploration into its existing achievements. It summarizes defects such as undesirable diversity of solutions to several multi-objective Particle Swarm Optimization algorithm, etc. Targeted for exiting defects and shortages in existing algorithms, it highlights settlement to the following problems: First, it explores dominating mechanism of multi-objective Particle Swarm Optimization algorithm, makes a comparative analysis between selection of non-dominating solution by traditional dominating mechanism and new dominating mechanism, and adopts traditional mechanism to select non-dominating solution to avoid loss of boundary point and maintain excellent distributivity to solutions. Second, during selection of global best selection, it puts forward the method to use spatial partition tree to index non-empty grid, define the conceptions of crowding distance and crowding distance density ratio, give preference to particles with the highest crowding distance density ratio as global best to make selection of global best more rational to solve the defect that distributivity of solutions is not good enough as a result of selection method of algorithm foundation by diversity of solutions based on saving non-dominated solution in Archive set. Last but not the least, in oil blending application, targeted for minimum cost and maximum profit, it establishes target functions during oil blending progress and apply the suggested algorithm in oil blending. Analysis on simulation experiment result verifies the feasibility of our algorithm suggested hereinto.As a new optimization technique, Particle Swarm Optimization algorithm is a new hot spot in the intelligent optimization field. Its application in solving multi-objective optimization algorithm enriches and develops the solving methods to multi-objective optimization and expands the application field of Particle Swarm Optimization algorithm.