Multi-Objective Particle Swarm Optmization Algorithm And Application

Particle swarm optimization is a good heuristic search technique to simulate the behavior and nature of a flock of birds which aim to find food. PSO has some attractive characteristics, such as versatility, simple in principle, easily implemented, cooperative search and so on. However, at the same time the algorithm is poor in local search ability, easy to fall into local minimum solution and defective in its theory. According to the shortcomings of PSO and the characteristics of multi-objective optimization, this paper puts forward some new mechanisms and strategies and improves the elementary particle swarm optimization algorithm. The main results of this research are as follows:(1) Research on how to use Pareto non-dominated theory, tournament selection and crowding distance method, improved the update strategy of particles, the aim is to solve multi-objective optimization problems and enhance the efficiency of algorithm.(2) Research on how to use two populations, immune mechanism and constrained non-dominated method, the aim is to solve constrained multi-objective optimization problems, improve the population diversity and avoid falling into local minimum.(3) Research on how to use the rate of crowding distance changing, adjust adaptively the performance parameter of the algorithm for improve the global search ability, constrainted optimization problems is convert reasonably, the aim to solve constrained optimization problems.(4) Research on how to make good use of mean strategy, improve traditional velocity update equation, the aim to avoid falling into local minimum in solving multi-objective optimization problems.(5) Research on how to use chaotic mutation operator, roulette selection strategy, select the elite from the optimal solution set, the aim is to avoid falling into the local minimum and solve non-uniform multi-objective optimization problems.