Research And Application Of Particle Swarm Optimization Algorithm

Optimization problem is an common question in engineering technology, scientificresearch and other fields. With the formation of computational complexity theory,people have found and proved lots of combinatorial optimization problem which iscalled NP and NP hard problems. Therefore, a series of modern optimization methodsemerge as the times require, such as simulated annealing, genetic algorithm, ant colonyoptimization algorithm. These algorithms have been successfully applied in some actualproblems.PSO algorithm has most advantages in artificial intelligence algorithms, such asthe simple calculation, fast convergence, robust, and so on. So PSO has became a hotspot in recent years. At the same time, the appropriate combination with geneticalgorithm, greedy algorithm, a particle swarm optimization algorithm local strategy, willbe greater to avoid local convergence, and then greatly improve the speed ofconvergence. So for practical problems, how to design the effective iterative strategy isthe key problem in particle swarm optimization algorithm.In this article we based on the particle swarm optimization algorithm, byintroducing the greed algorithm, genetic algorithm as its local strategy, and then somehybrid particle swarm optimization algorithms have been proposed. We applied thosealgorithm to solve the combinatorial optimization problem: the famous knapsackproblem, traveling salesman problem,2dimensional packing problems, respectively.The result of experiments shows that our algorithms are effective in those problem.Inthe third chapter, in view of the knapsack problem, the introduction of greedy algorithm,penalty function method is improved in the iteration problem in a large number of thefeasible solution of particle swarm optimization algorithm was designed, and thealgorithm for the simulation experiment, the analysis of experimental results werecompared. The traveling salesman problem and2d packing problem, we define a newmultiplication and addition operations, the particle swarm algorithm in the raw speediterative formula, using genetic algorithm to replace and improve, the hybrid genetic algorithm, particle swarm optimization algorithm is proposed. Particle swarmoptimization algorithm for traveling salesman problem the TSP,14of17node of theTSP and the Chinese postman problem three simulation experiments, and compares theexperiment result analysis. For the packing problem, we retain the genetic strategy forthe optimal solution, puts forward a hybrid particle swarm optimization algorithm withmemory. By introducing other intelligent algorithms are the most local strategy ofparticle swarm algorithm, can improve convergence efficiency in differentcircumstances, for processing the discrete situation more practical, but also can greatlyimprove the particle swarm optimization (pso) algorithm falls into local optimumsituation.