Font Size: a A A

The Research Of Improved Particle Swarm Optimization Algorithms For Solving Optimization Problem

Posted on:2018-10-19Degree:MasterType:Thesis
Country:ChinaCandidate:Q L ChenFull Text:PDF
GTID:2348330518979150Subject:Computational Mathematics
Abstract/Summary:PDF Full Text Request
Particle swarm optimization(PSO)becomes a hot research topic in the field of intelligent optimization because of its simple principle and implementation,fewer tuned parameters and satisfactory convergence performance,and it can effectively solve the discontinuous and non-differentiable optimization problems.In order to overcome the shortcomings of conventional particle swarm optimization(PSO)algorithm,such as slower convergence speed and lower search accuracy,and easily trapping in local optima.In this paper,the concept of the nearest equivalent particle and the optimal inverse direction is presented,and improve the algorithm for unconstrained continuous optimization problem,mixed integer problem and multi-objective optimization problem.Firstly,this paper briefly introduces the related concepts,research status and solving methods of optimization problems.Then we introduce the current swarm intelligence optimization algorithm and its research status,and the principle of particle swarm optimization algorithm is analyzed.Secondly,the different random replacement strategy is proposed based on the nearest equivalent particle of each particle.The algorithm used the simplified particle swarm to update particles,and it speed up the searching capability of the particles.In order to ensure the global search ability of the algorithm,we put forward the optimal random search strategy in the opposite direction for some particles with worst adaptive value.The algorithm tested on seven distinct types of benchmark functions.The results show that the convergence of the two methods makes the algorithm have a great improvement in the speed and accuracy.Thirdly,in view of the mixed integer optimization problem,the improved multidimensional inertia weights method is put forward,and each particle can have a choice to learn their ancestors' speed.At the same time,the linear decreasing step search strategy is used to balance the global and local search ability of integer variables.The chaos particle variation strategy is adopted on the quality of the worst part,so as toensure the population diversity.The experimental results verify the effectiveness of the improved strategy.Lastly,in view of the multi-objective optimization problem,the multidimensional inertia weights method is proposed to improve the performance of the algorithm.Simultaneously,the crowded entropy strategy is introduced to maintain and update the external files.The chaos particle variation strategy is adopted on the part of the particles which located far away from the optimal particle,so as to ensure the population diversity.The experimental results verify the effectiveness of the improved strategy.
Keywords/Search Tags:particle swarm optimization, unconstrained optimization problem, mixed integer optimization problem, multi-objective optimization problem
PDF Full Text Request
Related items