Fuzzy lifetime in genetic algorithm against premature convergence

Abstract

A large scale of design, control, scheduling or other engineering problems results in solution of optimization problems. In numerous areas of engineering there are problems to which Genetic Algorithms (GAs) can without difficulty be applied. Premature convergence is a classical problem in finding optimal solution in Genetic Algorithms. Varying the population size and the population diversity are the two possible ways of avoiding the premature convergence in a GA. If the population size is small, the diversity of population may be small and the GA will converge very quickly. On the other hand, if the population size is too big, the GA will take a lot of time to converge and this may cause wastage in computational resources. In this paper, we propose a Fuzzy Genetic Algorithm (FGA) which uses a new technique that is based on the lifetime for each chromosome. We use a conceptual of distance from the best chromosome in a population to create a fuzzy membership function which formed a new fitness function for the FGA. This new fitness function will then be used in a bi-linear allocation lifetime for varying the population size. Computational experiments are conducted to compare the performance of this new technique with some commonly used selection mechanisms found in a standard GA for solving some numerical functions from the literature.