Research On Simulated Annealing Hierarchic Genetic Algorithm And Its Application To (N+M) Fault-tolerant Systems

Genetic algorithm is a random searching method which simulates natural selectionand evolution. This method has some advantages that other usual methods don't havebecause of its two characteristics -implicit parallelism and global searching. However,simple genetic algorithm has defects of slow convergence and premature tendency. Sosome improved Genetic Algorithms are not only designed to eliminate its defects in thisthesis, but used to optimize (N+M) fault-tolerant systems as well.Firstly, the characteristic, the foundational theory, development and application ofgenetic algorithm are introduced, and the optimal model of (N+M) fault-tolerant systems isanalyzed in this thesis. Aiming to overcome the weakness of simple genetic algorithm, it iscombined with hierarchic method and simulated annealing thought so that three improvedgenetic algorithms are gotten. They are namely hierarchic genetic algorithm, simulatedannealing genetic algorithm and simulated annealing hierarchic genetic algorithm.On the one hand, the algorithms combine simulated annealing algorithm with geneticalgorithm in constraints, choosing of crossover and mutation probability, individualmutation and the like. On the other hand, the genetic algorithms are improved in severalaspects such as encoding, selection, crossover and mutation operator.Finally, the improved genetic algorithms, which are realized by Matlab language, areused to optimize (N+M) fault-tolerant systems, and its optimal solutions are obtained.Calculation result shows that the algorithms proposed in the paper are valid and correct.The performance analysis of the algorithms proves that its convergent speed are improved,and the better globally optimal solutions are achieved.