Research On Application For Hybrid Intelligence Algorithm

Swarm intelligence is a class of optimization algorithms form with groups of methods tomultiple solutions (also called individuals), some or all of the individuals through informationexchange to complete optimization algorithms, swarm intelligence optimization algorithmsare among the random nature search algorithms, including the most typical genetic algorithm(GA), particle swarm optimization (PSO), ant colony algorithm (ACO), etc., reflect thesuccess of these algorithms to simulate adaptive linear nature of biological importance inalgorithm design and viability. As the group of intelligent optimization algorithms can beobtained the optimal solution within a short time, gradually becoming a popular modernoptimization research.Particle swarm optimization (PSO) is a parallel global random search algorithm, it isconceptually simple and easy to achieve, the search speed range, and its basic idea is thatgroups of individuals through collaboration and information sharing in order to find out theoptimal solution. The current study focuses on PSO algorithm to improve the algorithms,mathematical analysis of algorithms, algorithm parameters, social and biological behavior,integration and comparison.Genetic algorithm (GA) based on biological genetic evolution, is the most widely usedintelligent optimization algorithm, involving many disciplines and can solve many practicalproblems.This paper presents a hybrid intelligent algorithm (HGC-PSO), PSO algorithm to solvethe premature convergence problem, the work was completed for two.The first job is the combination of two algorithms (PSO and GA), based on populationfitness variance to determine the mutation rate,1) Use Cauchy variation, except forpopulations in some of the best fitness value of particle algorithm to solve the Cauchymutation operation populations tend to process a single problem.2) Take Gaussian mutationinto consideration, changing the direction of the particle, in order to enhance the ability ofparticles out of the local optimum. This part is in Chapter4.The second job is the underwater environment modeling and analysis, and underwatervehicle navigation planning problem is transformed into three-dimensional space that, whilemodeling for the environment, the HGC-PSO algorithm is applied to underwater vehicle path planning in Chapter5.Premature convergence of PSO algorithm has been the focus of attention of researchers,taking into account the particle swarm algorithm is easy to fall into the latter part of theiterative local optimization, is not conducive to global search, and HGC-PSO introduces theGaussian and Cauchy mutation operator, the use of particle swarm convergence of thecalculation to determine the mechanism of mutation probability, a single search area to solvethe population into the local optimum problem. Experimental results show that, the searchtime of HGC-PSO algorithm is slightly longer than the standard particle swarm algorithmalthough, but it has good convergence properties, and effectively avoid the prematureconvergence.Underwater vehicle path planning is the core issue of the campaign to find a better route,researchers have proposed a variety of methods, such as human potential, segmentation, etc.;this paper is to try to HGC-PSO algorithm to underwater vehicle path planning, but only toexplore this issue with the HGC-PSO algorithm for planning feasible, and without muchconcern for the performance of its search results. From the experimental results can be known,HGC-PSO algorithm can be achieved essentially three-dimensional underwater navigationplanning, but the performance of its deep-seated to be further verified.