基于FVM-HGA的河流水质模型多参数识别

针对解析法求解河流水质模型的不足和传统方法在求解河流水质模型多参数识别问题遇到的困难,提出了采用有限体积法结合混合遗传算法(FVM-HGA)的参数识别算法。由于水质模型中源项的广泛存在,对流项的离散采用了满足有界性条件的非线性高精度格式-MINMOD格式。由于优化算法对参数识别反问题具有的普遍适用性,因而反演算法采用了全局寻优能力很强的遗传算法;为了克服传统遗传算法局部搜索能力较弱的缺点,采用了遗传算法与最速下降法相结合的混合遗传算法。两个算例的计算结果表明,采用FVM-HGA算法对常系数河流水质模型和变系数水质模型都能给出较好的反演结果。

Parameters identification of river water quality model based on finite volume method-hybrid genetic algorithm

For the difficulties of analytical method on solving river water quality model and traditional algorithms on solving the parameters identification problem of a river water quality model,a parameters identification method based on finite volume method-hybrid genetic algorithm(FVM-HGA) is proposed.The bounded nonlinear high resolution scheme-MINMOD is adopted in the discretization of the convective term because the source term exists widely.Since optimization algorithms could always be used to solve the parameters identification inverse problem,the genetic algorithm which has the excellent global search ability is adopted.The hybrid genetic algorithm combined with steepest method is adopted in order to avoid the shortcoming of weak local search ability of traditional genetic algorithm.Two computational results indicate that FVM-HGA algorithm can give a good precision inversion results for both the constant and variational coefficients water quality model.