A Research And Implement Of Genetic Algorithm

According their unmatched features, genetic algorithms have been closely concerned since its inception.1975, Holland enforces this effect by proposed the schemas theory. However, recent years many experts and scholars doubled the validity of schemas theory based on the existing of deceptive problem. This paper mainly researched these two issues and the main innovative conclusions as follow:1. In-depth researching the schemas theory under single-point-crossover. Because schemas theory has always been a core of genetic algorithm, in this paper we proposed many new definitions in allusion to the problem of definition overlapping between schemas survived and schemas created. Then deduced the conclusion about that which factors will affect the probability of schemas survived under single-point-crossover.2. Carefully analyzing the factors which caused deceptive problem. At present, deceptive problem is the most trouble issue of genetic algorithms. In this paper we carefully analyzed the basic cause of deceptive problem and then analyzed the reasons which could seriously affect the algorithm's convergence according to the appearance of deceptive problem, and brought forward some proposals to reduce its impact.3. Proposing MCAGA. MCAGA (Multi-factors Competition Adap-tive Genetic Algorithm) is proposed based on previous researches. Simulation results specified that MCAGA has largely enhanced the probability of global convergence when solving deceptive problems. At last, we used MCAGA which combined with the feature of problems to solve 0-1bags problem and acquired a good result.