Research On Several Intelligent Optimization Algorithms And Their Applications

The intelligence optimization algorithms and their applications have obtained widespread attention in recent years and in practice they have been applied in many fields. The differential evolution algorithm, the harmony search algorithm and the ant colony algorithm and their applications are studied in this paper. And the main work is as follows.(1) An improved differential evolution algorithm named KDE is developed to overcome the shortages of original differential evolution algorithm. In KDE, the entire population is grouped into serval subpopulations according to the individual differentiation information and different mutation operators are placed in different subpopulations. Based on the fitness of the population, it can regulate the mutation probability and crossover rate adaptively. The diversity of the population is maintained as much as possible depending on the grouping strategy and self-adaptive adjustment of the control parameters. Then the local search goes to the global search to avoid premature.(2) A modified non-dominated sorting genetic algorithm with elitist is raised and this algorithm is based on the Pareto optimality theory. The concept of dominance is used to evaluate and select the solutions to build the non-dominated solution set. A new grounding grid optimization design model is established through analysis and assumptions in electrical substations. The optimization of the grounding grid is conducted with the proposed algorithm to get the best design.(3) For the purpose of avoiding the disadvantages of the existed improved harmony search algorithms, an adaptive harmony search-particle swarm optimization algorithm is presented. In the harmony memory, each variable is updated using the PSO algorithm and adjusted by adaptive parameters PAR and bw to improve the search efficiency of multi-dimension problems. The improved algorithm is used to solve 0-1 knapsack problem and compared with basic harmony search algorithm and IHS algorithm. The simulation results show that the FAHS algorithm is more efficient than IHS and HS.(4) A new multiple attribute decision-making method based on fuzzy expert information is proposed in this paper. Each attribute is fuzzily sorted by expert information and then the weights are determined objectively and the optimization decision is made by the projection pursuit model. The newly proposed model can both balance the experts' preference to the attribute and the objectivity of the weight determination which makes the decision more reasonable and reliable. Meanwhile an adaptive clustering differential evolution algorithm is applied to optimize the decision-making model and the testing result shows that this algorithm can improve optimization searching efficiency and the ability of jumping out the local optimums.(5) An improved ant colony algorithm with retracing is proposed in this paper to solve the problem of dead end in the search, and the convergence speed of the ant colony algorithm is improved. The automobile navigation problem is studied in this paper and a strategy based on dynamic limited regional searching is introduced upon the commonly used algorithms. The path is optimized twice successively by A* algorithm and ant colony algorithm to improve the searching efficiency and precision. The simulation result verifies the efficiency of the algorithm in this paper.