Research And Application Of Barebone Multi-Objective Particle Swarm Optimization Algorithm

The optimal solution of single-objective optimization problem is usually unique,and in many practical problems such as vehicle path planning,power resource scheduling,workshop management scheduling,etc.,there is more than one objective that needs to be optimized,such problems are multi-objective optimization algorithm,and the solution of multi-objective optimization problem is usually a Pareto set.Multi-objective evolutionary algorithm is a common method to solve multi-objective optimization problems.The current multi-objective evolutionary algorithm still has some shortcomings in the convergence speed and the ability to solve the high-dimensional multi-objective optimization problem.Improving the performance of multi-objective evolutionary algorithm is a hot and difficult topic.Firstly,the thesis introduces the mathematical definition,research history and present situation,main methods and advantages and disadvantages of multi-objective optimization problem.Then,several classical multi-objective evolutionary algorithms,performance evaluation indexes and test function sets are introduced in detail.Then a barebone multiobjective particle swarm optimization algorithm based on the direction vector(DV-MOPSO)is proposed.This algorithm uses an improved direction vector to decompose the multi-objective optimization problem,which accelerates the convergence speed of the algorithm and enhances the ability of the algorithm to jump out of the local optimum.By comparing with several popular multi-objective optimization algorithms on the ZDT test function,the experimental results show that the convergence speed and convergence of the DV-MOPSO is better than that of other popular multi-objective optimization algorithms.In order to better solve the highdimensional multi-objective optimization problem with the number of objectives greater than 4,the thesis proposes a decomposition based multi-objective barebone particle swarm optimization algorithm(BB-MOPSO/D),and introduces a faster convergence barebone particle swarm optimization algorithm in the BB-MOPSO/D algorithm from the perspectives of convergence and diversity.Then,a new penalty boundary function is proposed,which increases the selection pressure of the diversity of the archive set and reduces the convergence loss as much as possible.Through on DTLZ test function and several popular multi-objective optimization algorithm has carried on the contrast experiment,the results show that BBMOPSO/D algorithm is better than that of several other popular multi-objective optimization algorithm,especially when the objective number increased to 10,significantly better than several other algorithms,that BB-MOPSO/D algorithm can better solve high-dimensional multi-objective optimization problem.The thesis applies BB-MOPSO/D algorithm to the optimization of mobile sink path in wireless sensor network,and verifies that the BBMOPSO/D algorithm can solve the optimization of mobile sink path well through simulation experiments.In view of the more and more complex multi-objective optimization problems,the two multi-objective barebone particle swarm optimization algorithms proposed in the thesis have shown better comprehensive performance in dealing with two-objective optimization problems and high-dimensional multi-objective optimization problems,and have certain application value.