Research On Particle Swarm Optimization Algorithm Based On Intermediate Structure Layer

The formation of swarm intelligence algorithm provides new ideas for solving optimization problems,and also solves many practical problems in the field of engineering technology.Particle swarm optimization algorithm is a kind of swarm intelligence algorithm.Because of its advantages of high efficiency,less parameters,and easy to implement widely attention,the PSO is widely watched in the field of swarm intelligence algorithm.Structure of middle layer is a method in the theory of complex systems,which makes the system hierarchical.It is convenient for the researchers to observe and study the different components of the system independently,which belongs to a kind of modularity and damage control strategy widely used in theoretical disciplines.Because the PSO belongs to group intelligence algorithm,and cluster intelligence is derived from the simulation of real complex system,therefore,based on the PSO introduced in middle layer structure,so as to make it more close to the real complex system and enhance the intelligent emergence of the system,comprehensively improve the overall performance of the algorithm.The main work of this article is as follows:1.A hierarchical particle swarm algorithm(LPSO)based on intermediate structure layer is proposed based on the analysis and research of particle swarm optimization and complex system theory.By introducing the concept of "group",an intermediate layer composed of different groups is constructed.And according to the type of group definition of the corresponding search strategy.The experiment shows that this method improves the problem of the standard PSO structure and the loss of particle diversity,and effectively inhibits the premature of the algorithm.2.In order to further enrich the intermediate structure layer and enhance the complex system characteristics of particle swarm system,a hierarchical particle swarm optimization(GLPSO)algorithm based on evolutionary mechanism is proposed based on LPSO algorithm.The algorithm is added in the middle structure layer mutation and cross operation,in which the mutation operation to enhance the diversity of particles,is conducive to the particles out of the local extreme value,to expand the search range of particles,and crossover operation is helpful to improve the quality of particles,in order to gain better convergence speed and precision.Finally,the experiment proves that this strategy greatly improves the comprehensive performance of the algorithm.3.Traveling salesman problem is one of the typical combinatorial optimization problems.But the research and application about the discrete particle swarm algorithm is very immature,especially using the particle swarm algorithm to solve traveling salesman problem is a new research direction.Therefore,based on the GLPSO algorithm,this paper designs a discrete PSO to solve the TSP problem,and tests the algorithm with the classical test case in the TSPLIB standard library.The test results show that the algorithm is more efficient,stable and practical.