Improved Particle Swarm Optimization Algorithm Based On Genetic Thinking And Applied Research

Particle swarm optimization (PSO) is swarm intelligence evolutionarycomputation methods with the global strategy and inspired nature, the basic ideaderived from biological simulation from the foraging activities on birdpopulations, proposed jointly by the American psychologist Kennedy andelectrical engineer Ebethart in1995,is an emerging based on swarm intelligencetheory of computing technology. The algorithm mainly through competition andcooperation mechanisms optimization guidance in the population of particles inthe searching process, and has good versatility and global. PSO algorithm hasbecome a typical representative of the swarm intelligence, and because of itsefficient global optimization result was a wide range of applications. Applicableto a variety of complex optimization problems and combinatorial optimizationproblems, the method is simple and easy to operate, to achieve fast optimizationconvergence, has been an important branch of research in artificial intelligenceexperts and scholars over the past decade.Find the ways to improve for easy to premature convergence, the searchaccuracy is not high in the iteration of the late efficiency, easy to fall into localminima advantages ofe the algorithm defects, and applied more expansion areas,has very important oretical value and practical significance. This paper proposedimprovement program and application experiments based on the analyzes aboutthe basic principles and development of the PSO algorithm. Research work andinnovations can be summarized as follows:First, proposed an improved particle swarm optimization algorithm(GAPSO).In the research, analyzed the characteristics and principles of swarmintelligence algorithm and compare particle swarm optimization algorithm withother algorithms, the obtained algorithm integration is more feasible ways toimprove it. In terms of the nature and characteristics of Genetic algorithm andparticle swarm optimization algorithm can produce a good match, the operator of the genetic operation is introduced into the iterative process of the PSO algorithmwith the appropriate selection, crossover and mutation operators, and take theinertia weight factor nonlinear decreasing mechanism, resulting in improvedparticle swarm optimization algorithm based on genetic operators and nonlineardecreasing inertia weight factor. The last four standard test functions, theimproved algorithm has a good method to optimize the performance verification,simulation results show that the improved optimization algorithm not onlyeffectively overcome the inherent defects of the traditional PSO algorithm, butalso improve the search accuracy of the algorithm to avoid prematureconvergence to local minima phenomena.Secondly, use of the improved optimization algorithm (GAPSO) to optimizethe PID controller parameters. The settings of the PID controller parameters candirectly affect the control performance of the controller, is ultimately based onerror feedback to eliminate or minimize the error of the control strategy, thecontrol principle is simple, easy to operate to achieve, and have a higherreliability. Tuning parameters related to the core of the entire PID controlapplication, in order to a good parameter optimize results the improved particleswarm algorithm (GAPSO) will be into the application of PID parameteradjustment and optimization tuning, and simulation results prove that thealgorithm has the better parameters optimization effect.Finally, genetic particle swarm optimization (GAPSO-BC) based on beeswarm strategy and inertia factor decreasing to optimize the training of the neuralnetwork. Optimize the training method proposed for the verification of thischapter to be effective, will be introduced in practical applications such as intothe classification and function graph approximation, simulation results show thatthe optimization algorithm can overcome the various problems of he traditionalneural network, the generalization ability of neural network to be enhanced,while also improving learning ability and convergence speed of training.