Research And Analysis Based On Improved Multi-objective Optimization Problem Using Particle Swarm Optimization Algorithm

In real life and engineering problems, in general, will encounter a lot of multi-objective optimization problem, there are some drawbacks and defects when traditional solution be used. The particle swarm algorithm has fast convergence and parameter setting and simple, it is widely utilized to solve the multi-objective optimization problem. The multi-objective particle swarm optimization algorithm theoretical system is not perfect, and multi-objective particle swarm optimization algorithm has easy to fall into premature convergence, accuracy is not high shortcomings reconciliation set unevenly distributed. This paper will focus on multi-objective particle swarm optimization algorithm corresponding improvement measures and strategies, and the robot path planning application improved algorithm. The main research works are as follows:(1) For the ability of the algorithm is easy to fall into local and precocious puberty jumping out of the problem is not strong, this paper proposes a Gaussian perturbation strategy to make the particle swarm algorithm in solving multi-objective optimization problem to enhance the ability to jump out of local optima effectively avoid the algorithm into premature. Combined with the inertia weight particle swarm velocity update formula, given the realization of Gaussian disturbances policy approach.(2) Proposed a judgment particle current state of convergence approach. Particle Swarm in early and late of the evolution of each in a different state of convergence, a single parameter selection group easily lead to the evolution of efficiency is not high. Judging by the iterative algorithm to generate new non-dominated solutions in an external file to determine the situation of the dominant state of convergence in the particle current, adjustable parameters to improve the efficiency and accuracy of population evolution.(3) Convergence of the questions of multi-objective particle swarm optimization algorithm accuracy is not high, and uneven distribution of the solution space, proposed based on particle converged state of inertia weight adaptive strategy. This strategy can improve the convergence speed and convergence accuracy.(4) In order to maintain and improve the convergence speed, add the external file simplex crossover operator, the paper is improved multi-objective particle swarm algorithm with a certain direction, so as to improve the convergence speed and accuracy of the algorithm purposes.(5) Improved multi-objective particle swarm algorithm made application to the robot path planning,modeled robot task, and then make an objective function, multi-objective optimization particle swarm algorithm for optimal path through the basic multi-objective particle swarm optimization algorithm simulation comparison.In this paper, the standard multi-objective optimization test function set ZDT series and DTLZ system algorithm experiments, algorithms and basic MOPSO improved algorithm and multi-objective evolutionary algorithm NSGA-II were compared, the experimental results show that the proposed algorithm access to better Pareto front can be obtained more uniform and accurate, non-dominated solutions. Robot path planning problem, through simulation experiments in two different environments, and standard multi-objective particle swarm optimization algorithm comparison can show that the improved algorithm is more enough to get a better path.