Abstract: Selection operator is one of the most important operators for genetic algorithms. It has significant influences on the population’s diversity and the convergent speed of the algorithm. This paper improves the roulette wheel selection operator, applies partitioning of fitness to individual selection of the population, proposes roulette wheel selection based on partitioning of fitness, and applies the approach to the optimization of the Poisson curve settlement forecasting model. The experimental results indicate that, this algorithm provides an effective approach for the parameter estimation of the Poisson curve settlement forecasting model.

Key words: selection operator, genetic algorithm, Poisson curve, parameter estimation