Memory Genetic Algorithms In The Application Of Function Optimization

Genetic Algorithm is a reference biological and genetic mechanisms of natural selection highly parallel, randomized, adaptive search algorithm for global optimization probability. Attach great importance to domestic and international theory and application of genetic algorithm research, and has made remarkable progress. However, theory and method of genetic algorithms are not yet mature, the algorithm itself to a number of shortcomings remains to be further improved and perfected, and function of genetic algorithm optimization is used for performance evaluation example. In the genetic algorithm function optimization process, the emerging phenomenon of the individual repeat, repeat probability and parameter settings. A large number of individuals in each generation of the duplication of genetic evolution, in the calculation of the fitness of each individual, the same individual, the same operation, the same results, consume a lot of run time and improve the time complexity of the program for This phenomenon, this paper presents a new algorithm - Memory genetic algorithm. The creation of a genetic algorithm is the concept of the individual library, individual library to save the genetic code of individuals and fitness information, the same as this improved method avoids the double counting of individual fitness, fitness evaluation can greatly reduce the time to get faster convergence speed, improved performance of the algorithm. Study functions of one variable from the simple, binary multimodal function and single-peak function of the iterative process carried out intensive exploration, select some typical simulation test of complex functions, by comparison with other methods show that the genetic memory of the proposed algorithm can effectively speed up the convergence rate,thus improving the performance of the algorithm.In this paper, genetic algorithm and the key improvements are as follows:1. Genetic algorithms simulate the natural selection and genetic replication occurs, the phenomenon of crossover and mutation, starting from any initial population by random selection, crossover and mutation, resulting in a group of individuals better adapted to the environment. Firstly, on a function optimization details of the genetic algorithm the actual work process, in the process discovered the individual duplication; then a simple function of one variable specific analysis of the genetic algorithms process populations of individuals overlap, including what parameter settings affect repetition rate, based on studies of these phenomena made the subject provided a factual basis.2. Operation according to the standard genetic algorithm process, the implementation of each individual in each generation of selection, crossover and mutation operations before, are to conduct the calculation of fitness value. Obtained under the previous instance of each generation have a higher repetition rate, this paper proposes the concept of the genetic individual library, individual library code stored in the individual and the fitness value of information, resulting in the implementation of the new individual selection, crossover and mutation before first with the individual library of individual coding information has been there for comparison, if the individual library has the same encoding of individual information, directly out the corresponding fitness value, to avoid the same fitness value of individual double-counting, speed up the speed, improved performance of the algorithm.3. To test the validity of the proposed improvements and whether certain versatility, choose a number of test functions, using a variety of algorithms were tested through a large number of experiments and the analysis and comparison of experimental results, gives the conclusion.Finally, a summary of the research, pointed out that this modified algorithm the advantages and disadvantages of further study of genetic algorithm to provide a reference value.