Improvement And Application Of Multi-objective Particle Swarm Optimization Algorithm

Multi-objective optimization problems exist widely in engineering design,manufacturing,information processing and other applications.Particle swarm optimization algorithm is characterized by fast convergence,simplicity and parallel search,and is especially suitable for solving multi-objective optimization problems.In this paper,the multi-objective particle swarm optimization algorithm is systematically studied,including the following:(1)Aiming at the problem that the particle swarm optimization algorithm is sensitive to the value of control parameters,a multi-objective particle swarm optimization algorithm with fewer control parameters is proposed.In order to enhance the global search ability of the algorithm,the particle with quantum behavior is used to search the solution space.In order to prevent the population from premature convergence,a mutation operator which changes the range of action with time is introduced.In order to improve the diversity of particle swarm,an external archive is used to preserve the Pareto optimal solution.The self-adaptive grid method is used to update and maintain the external archive,and then the global optimal leader particles are selected from them,so that the population can quickly approach the real Pareto optimal frontier and maintain good distribution characteristics.Compared with three existing algorithms,the proposed algorithm has better convergence and distribution(2)In order to solve the problem that traditional multi-objective optimization algorithms are difficult to retain multiple Pareto optimal solution sets,a multi-modal multi-objective particle swarm optimization algorithm(RNMOPSO)using ring topology and neighborhood perturbation strategy is proposed.Using the index based non overlapping ring topology structure,the population can form multiple independent search niches without specifying any niche parameters,so that the algorithm can search more optimal solutions.An automatic conversion mechanism between global search and local search is proposed to better balance convergence and diversity.The stagnation detection strategy is introduced to disturb the neighborhood optimal particles,so as to improve the diversity of particle swarm optimization and prevent the algorithm from premature convergence to a Pareto optimal solution set.The simulation results of 11 multi-modal multi-objective test functions show that the proposed algorithm can search more complete and evenly distributed Pareto optimal solutions while ensuring the diversity and convergence in the objective space(3)A multi-objective particle swarm optimization method for power system scheduling problem is studied.In this paper,a mathematical model is established for the environmental economic dispatch problem considering the generation cost and pollution emission.The particle swarm optimization algorithm with fewer control parameters is used to solve the problem.The integer coding method is used to establish a suitable mapping relationship between the position of the particle in the search space and the solution of the environmental economic dispatch problem.An effective method is adopted to solve the problem The non feasible solution of the environmental economic dispatch problem is adjusted quickly by the constraint processing method of equality constraint,and the compromise solution of the environmental economic dispatch problem is extracted by introducing the fuzzy set theory.The improved algorithm is applied to IEEE-30 bus test system to verify the feasibility and effectiveness of the designed algorithm in dealing with environmental economic scheduling problems.(4)This paper studies the multi-objective particle swarm optimization method of feature selection problem,analyzes the multi-modal characteristics of the actual feature selection problem,and uses the multi-modal multi-objective particle swarm optimization algorithm proposed in this paper to solve the problem.The algorithm is applied to seven datasets,and the experimental results show that the algorithm can obtain multiple sets of equivalent feature subsets on the basis of ensuring the final classification accuracy,and reduce the cost of feature extraction...