Research On Hunter Particle Swarm Optimization With Backtracking Mechanism

With the rapid development of modern science and technology, continuousimprovement of people’s living standards, expanding the range of human activities, therapid depletion of natural resources, the people on the existing technology put forwardhigher requirements, to more efficient production activities, which optimizationalgorithms and optimization techniques are an important part. Particle swarm optimizationalgorithm based on swarm intelligence is to calculate the evolution of an emergingtechnology proposed by Kennedy and Eberhart in1995. Particle swarm optimization in thehandling of some of the convergence effect unimodal function well, but with theexpansion of the scale of the problem, the algorithm appears difficult to converge or not toconverge, prematurity and other shortcomings. Aiming at these problems particle swarmalgorithm to establish a new processing model, and gives a solution.First, this paper proposes a new particle swarm model-Hunter Particle SwarmOptimization. This model is to make the particles have a higher intelligence of birdsforaging in the original simple model based on the particle not only has the ability to knowhow far away from their own food, but also found that the position of those predators, andafter flying in initiative avoid these predators. In this mechanism, we use the steady-statecriterion brief analysis of discrete systems predatory convergence of particle swarmalgorithm and particle swarm algorithm implementation predators, the other particleswarm algorithm, particle swarm compared with predatory greater convergence accuracy.Secondly, for the shortcomings of predatory PSO algorithm, such as the poorefficiency of the algorithm, the algorithm takes called the disadvantages presentedpredatory particle swarm backtracking mechanism. On the basis of the theoreticalframework of particle swarm prey on the proposed use of backtracking mechanism toretain as much as possible the results of the particle swarm computing, based on theoriginal results of the search continues, and avoid known predators. Through the sameprocess analysis algorithm gives a detailed analysis of convergence. Finally algorithm.Compared to the predatory swarm, backtracking mechanism of particle swarm algorithm greatly reduces time-consuming.Finally, a numerical comparison with the existing variety of new particle swarm, indealing with multimodal functions, this algorithm has a good effect.