中文摘要：

为了克服传统参数识别方法在进行河流水质模型多参数识别时的不足,以一维纵向离散方程为基础,构建了同时确定河流水质模型水力和水质参数的反演优化通用算法。利用反演优化法重点探讨了初值选取、观测噪声、测量精度等因素对参数计算结果的影响,比较了变尺度法和单纯形法两种优化算法的反演效果。算例计算结果表明,单纯形法在可靠性与精度方面均要优于变尺度法。结合瞬时源和连续源两个算例验证了该方法的可靠性。两个算例的计算结果表明,采用反演优化算法对瞬时源河流水质模型和连续源水质模型都能给出较好的反演结果。

英文摘要：

A general inversion-optimization algorithm was constructed for determination of hydraulic and water quality parameters of a water quality model such as longitudinal dispersion coefficient and cross-sectional average velocity to overcome the difficulties of traditional algorithms in parameters identification.It was used to examine in details the influences of observation noise,instrument precision and initial value on estimation of parameters.A comparison was made between BFGS quasi-Newton algorithm and Nelder-Mead simplex algorithm via case studies and the results indicate superiority of the former in stability and accuracy.The new algorithm gives good accuracy of inversion results in the water quality models of instantaneous source and successive source.