Rank-based varying population size genetic algorithm with sizing strategy

This paper presents a rank-based varying population size genetic algorithm. A sizing strategy is proposed to determine the increased population size. The use of rank-based fitness increases the population diversity, and the concept of age and lifetime save computation resources. According to the feasibility, the initial population is chosen using a reinitialization process. Offspring are generated by means of rank-based reproduction, parametric crossover and bit mutation in the process of evolution. Finally, numeric results of a pendulum-cart system control are obtained to demonstrate the developed genetic algorithm.