An Improved Genetic Algorithm for Parameter Calibration of Xin’anjiang Model

Abstract

Based on the analysis of Xin’anjiang model parameters, a multi-objective parameter optimization model is established. An improved genetic algorithm is proposed to solve inferior local searchability, premature convergence and slow convergence speed which are the defects of genetic algorithm. The suggested algorithm narrows the range of parameter constraints and at the same time makes use of direct comparison-proportional method and niche operator to achieve step-by-step parameter optimization. The improved genetic algorithm is successfully applied to parameter calibration of the Xin’anjiang model for the Zhuganhe water shed. The better parameter results obtained from the algorithm meet the accuracy requirements in calibration and verification. The calculated results and a comparison with Traditional Genetic Algorithm and SCE-UA algorithm show that the proposed IGA has not only higher rate of convergence, but also better optimization results. It provides an efficient and reliable optimization algorithm for the calibration of Xin’anjiang model parameter.