Equal-Width Partitioning Roulette Wheel Selection in Genetic Algorithm

Selection operator is one important operator in genetic algorithm (termed GA). It has significant influences on the performance of algorithm. Roulette wheel selection is a frequently used selection operator in implementation of GA. However it does not perform sufficiently well in balancing the convergence speed and population diversity of the algorithm. This paper proposes a novel roulette wheel selection based on fitness equal-width partitioning. The proposed selection operator groups the individuals by equal-width partitioning of the fitness interval of the whole population. And then in each time of selecting an individual to generate the new population, a group of individuals is selected with the method of roulette wheel selection, where an individual will be then chosen for survival in the new population. By restricting the fast reproduction of the majority of individuals sharing similar fitness, the proposed selection operator can sustain the population diversity to avoid premature. Encouraging experimental results demonstrate that the proposed selection operator is able to achieve better solution and has a faster convergence speed, compared to the traditional roulette wheel selection.