The Research And Application Of Genetic Algorithm

Genetic algorithm is a highly parallel,randomized and adaptive search algorithm, developmenting on natural selection and evolutionary of biological mechanisms. Its research history is short, early it is a constructed artificial system model of simulating the mechanisms of biological evolution from attempting to explain the complexity process of biological in the natural system. Formed in recent years around the world boom of evolutionary computation, computational intelligence became one of the major directions of artificial intelligence research, as well as the subsequent rise of artificial life research, so that genetic algorithms is concerned broad.Genetic algorithms in view of the existence of slow convergence or prone to "premature" phenomenon, this paper improved genetic algorithms and advanced greedy 3PM crossover operator for TSP problems, at the same time improved the simulated annealing algorithm and two improved algorithms combination ,formed Greedy 3PM Crossover operator Based Simulated Annealing Genetic Algorithms(GCBSAGA). The main research work is listed as follows: Firstly, introduced genetic algorithm research at home and abroad, as well as the existing genetic algorithm and simulated annealing algorithm to combine the existing situation. Secondly, the genetic algorithm in the basic concepts, principles, methods, steps, based on the tour for the traveling salesman problem (TSP), advance Greedy 3PM Crossover operator Based Simulated Annealing Genetic Algorithms(GCBSAGA). In the new algorithm, on the one hand, the improved genetic algorithm, first to improve its framework, and then search to determine the efficiency of the introduction of the conditions, once the search efficiency, then the termination of genetic algorithm, so the genetic algorithm to retain the overall costeffective search, reject the latter part of the problem of low search capabilities. Then goes on to suggest a new crossover operator—greedy 3PM crossover operator, the operator will not only take into account the relationship between the city locations, more importantly, takes into account the relationship between cities edge. The method using the three parents produce offspring, can be better side information, and the use of bidirectional rotation of the ways in which subcompletely inherited parent information edge. But also to the offspring resulting from annealing was also carried out, thus the temperature of the concept will be added to the genetic algorithm, so that does not appear to make the basic offspring degradation. Can be seen through the experiment, greedy crossover operator 3PM very fast convergence, high search capabilities. On the other hand, simulated annealing algorithm, which uses genetic algorithm the initial solution is obtained after the better overall solution, that is better overall solution to conduct an independent annealing, which makes the blind search of the original simulated annealing algorithm fixed promising search area has greatly enhanced the efficiency of the search. Simulated annealing algorithm is used in genetic algorithm initial temperature after the end of the temperature, so that the use of high-temperature genetic algorithm, to play its highly efficient global search capability, the use of low-temperature simulated annealing algorithm, to play its strong local search ability. And then selected the internationally recognized TSP in the database for testing and found the TSP library than the optimal solution to provide better solutions, proved that the new algorithm. Finally, the new algorithm applied to the integration of distribution and recovery of the vehicle routing problem of the practical problems, and adding a codec mechanism for the specificity of the problem, through experiments and contrast GCBSAGA algorithm verified to solve such problems in the efficiency.