Improvement Of Particle Swarm Optimization Algorithm And Its Application In Parameter Optimization And Community Discovery

Particle Swarm Optimization(PSO)is an algorithm based on artificial life and evolutionary computation theory.More and more scholars focus on the theory Based on its advantages including simple algorithm easy to implement,fast convergence speed and few adjustable parameters,etc.Theoretical research and engineering applications continue to develop into a large number of areas.However,it is obvious that the traditional PSO algorithm has the problems of easy to be premature,slow convergence speed,and poor treatment of discrete optimization.Therefore,it has important theoretical value and practical significance to study the improvement and application of PSO.Aiming at the shortcoming that the PSO is easy to fall into the local optimum and the slow convergence rate,this thesis proposes a Hybrid Clonal Selection Particle Swarm Optimization(HCSPSO)algorithm which is introduce clonal selection strategy into the PSO.This algorithm consists of individual extreme value of the individual particles generated by the particle swarm algorithm to form a temporary clonal population;The Cauchy variation of the individual in the population is used,in order to increase the diversity of the population and improve the global search ability of the algorithm.Standard function test results show that HCSPSO has the advantages of fast convergence,strong global searching ability and high accuracy.The simulation experiment results show that the HCSPSO-PID controller can get better control effect with the parameter optimization of PID controller.In this thesis,an improved PSO(Local Iterative Particle Swarm Optimization,LIPSO)is proposed to detect the community structure of complex networks under the condition that the network cluster structure is very fuzzy.In this algorithm,the population is initialized with the idea of label propagation,and then,according to the state of the population and individual extreme value,using the method of comparison and mutation to update the speed of V.Then,according to the value of V,the particle in the population is updated by local iterative method.Finally,the individual and individual extremum in the population will be operated by a single crossover operator.Test results show that the structure of network clusters generated by the algorithm in the computer extremely vague compared to correlation algorithm has better clustering accuracy,and in real world network is able to dig out the community structure of higher quality.In this thesis,the principle of PSO algorithm and the characteristics of each parameter are analyzed theoretically,and two improved PSO algorithm are proposed.And they are applied to the parameter optimization of PID controller and the discovery of complex networks,promoting theoretical research and engineering application of PSO algorithm.