A Genetic Algorithm with Elite Crossover and Dynastic Change Strategies

Abstract

This paper proposes an elite crossover strategy together with a dynastic change strategy for genetic algorithms. These strategies are applied to the elites, with a different crossover operation applied to the general population. This multi-crossover operation approach is different from the traditional genetic algorithms where the same crossover strategy is used on both elites and general population. The advantage of adopting a multi-crossover operation approach is faster convergence. Additionally, by adopting a dynastic change strategy in the elite crossover operation, the problem of premature convergence does not need to be actively corrected. The inspiration for the dynastic change strategy comes from ancient Chinese history where royal members of a dynasty undertake intermarriages with other royal members in order to enhance their ascendancy. The central thesis of our elite crossover strategy is that a dynasty can never be sustained forever in a society that changes continuously with its environment. A set of 8 benchmark functions is selected to investigate the effectiveness and efficiency of the proposed genetic algorithm.