Optimization Algorithm, Based On Discrete Multi-objective Particle Swarm Optimization

Many practical application problems are the complex optimized questions, which belongs to the multi-objective optimization question and has many objective function. There are mutual connection and conflicts between the many objective function. It is unable to obtains a global optimal solution as the single objective optimization, which causes many objective to achieve optimal solution simultaneously. What can obtain is only a group of non-dominated solution, namely Pareto optimal set. Finally multi-objective optimization need to select a best trade-off solution from the Pareto optimal set, therefore multi-objective optimization problem become a multi-objective decision making problem in the end.At the same time, many practical application question belongs to the discrete variable optimization, or the continuous variable and the discrete variable simultaneously existing, and the pure continuous variable optimization are not many. Now there are many optimization method, but most aims at the continuous variable optimization and few on discrete variable optimization, even less on the discrete multi-objective optimization problems. The existing optimization of discrete variables mostly deal with the discrete variable using optimization method of the continuous variables, which will cause the big deviation and sometimes can not be able to obtain the solution. As one kind of efficient parallel optimization algorithm, the paticle swarm optimization is simple and convenient, easy to operate.Since the day it is proposed, particle swarm optimization receives the numerous researchers' attention, which has been made many kinds of improvements and widely applied in many kinds of practical optimization problems. But most of the existing optimization algorithms are suitable for the optimization of continuous variable. As result of its own iterative mechanism, particle swarm optimization cannot be applied to the discrete variable optimization directly and need some improvement before being used.This article carry on the analysis and classification to the existing algorithm used in the discrete multi-objective optimization, based on the principle, parameter choosing and aplication to the multi-objective optimization problem of the particle swarm optimization, then propose an discrete multi-objective optimization based on the particle swarm optimization. This algorithm adopt the binary mechanism to discretize the position vector of the particle and simultaneously establishe the non-dominated solution set to deposit the non-control solution collection searched, which increases the multiplicity of the non-control solutions. This article utilizes multi-objective trial function SCH, ZDT1 and ZDT3 to carry on the test to this optimized algorithm and its performance is good when compared with the other optimized method; And utilizes it to solveing the resource distribution problem, then obtaining the superior non-control solution. Finally, the article carries on the summary to the paper research, and look forward to the the development trend of the discrete multi-objective optimization methods in the future.