Parameter estimation of P-III distribution based on GA using rejection and interpolation mechanism

This paper studied the genetic algorithm (GA) deeply for the problem of hydrological frequency parameter estimation of P-III distribution. For simple GA, the convergence speed and global searching ability are contradictory. An improved algorithm named GA using rejection and interpolation mechanism (GA-RIM) are proposed. Firstly, the rejection mechanism is adopted to ensure the diversity of population and avoid useless operations. Secondly, the strategy of preserving excellent individual is used to guarantee the population to converge to the optimal solution. Thirdly, interpolation mechanism is used to ensure the population to explore in the total domain of definition and mutate adaptively according the diversity of population. The GA-RIM, GA and other four usual methods are used to estimate hydrological parameter of two examples. Through simulation experiments, it was found that the GA-RIM is superior than GA and other methods in terms of convergence speed, precision and global searching ability.