Dam & Safety ›› 2019, Vol. 0 ›› Issue (2): 16-.

Previous Articles     Next Articles

Calibration by system response for prediction model parameters and application test in Jinping hydropower station

LUO Hao, DAI Xiao-long, SHAO Peng-zhe and ZHANG Bo   

  1. Yalong River Hydropower Development Co., Ltd.
  • Received:2019-01-02 Online:2019-04-08 Published:2019-04-08

预报模型参数系统响应率定方法及锦屏水电站应用检验

罗  浩,代小龙,邵鹏哲,张 波   

  1. 雅砻江流域水电开发有限公司,四川 成都,610051
  • 作者简介:罗 浩(1989- ),湖南长沙人,工程师,主要从事水工安全管理工作。

Abstract:

For conceptual hydrological model for river basin, its core is the model structure and parameters. In general, the model application results and the forecasting accuracy largely depends on the model parameters. Therefore, to find a proper parameter optimization method to obtain optimal parameter values is essential. Usually, sum of squared errors is taken as the objective function to get nonlinear function parameters, which give rise to problems about irrelevant local optimal solution. Therefore, Professor BAO Weimin proposes the calibrating by system response rate method for model parameters. This method is used in parameters optimization of Xin'anjiang model and the ideal model as well as the practical basin model are established to verify the applicability and effectiveness of the method. Firstly, ideal model of secondary flood is built to verify its calibration and optimization ability for sensitive parameters. The result proved the applicability of the method as the convergence to truth-value is achieved and initial values exert little influence on result. Secondly, this method is applied to the actual river basin. Select the Liewa-Jinping reach and Maidilong-Jinping reach as the subjects, calibration of sensitive parameters in sub-model is carried out. The results show that the calibrated parameters are independent from the selection of initial value and the optimal value could be found. Moreover, with a relatively less times of optimization and higher accuracy, the effectiveness of calibration by system response for prediction model parameters is proved.

Key words: parameter calibration, Xin'anjiang model, ideal model, practical basin model, Jinping hydropower station, application tests

摘要: 对于概念性流域水文模型,其核心是模型的结构和参数。一般而言,模型的应用效果和预报的精度很大程度取决于模型参数。因此,寻找一种恰当的参数优选方法,使模型参数达到最优值至关重要。鉴于常规将误差平方和看作目标函数求解非线性函数参数,容易增加无关局部优值解问题,包为民教授提出了模型参数系统响应率定方法。笔者将模型参数系统响应率定方法应用于新安江模型参数优选中,建立了新安江模型的参数系统响应率定方法,并构建了理想模型和实际模型,以验证该方法的可行性和有效性。首先,构建次洪理想模型,用理想模型验证参数系统响应率定方法优选敏感参数的能力,结果表明,均能收敛到真值附近,且受初值的影响小,验证了该方法的可行性。其次,将该方法应用到实际流域。选取锦屏电站列瓦、麦地龙-锦屏区间,对流域中次模型敏感参数进行率定。结果表明,率定的参数与初值选取无关,基本能搜寻到最优值,寻优次数相对较少,精度相对较高,由此,说明了参数系统响应率定方法在实际流域中具有一定的有效性。

关键词: 参数率定, 新安江模型, 理想模型, 实际流域模型, 锦屏水电站, 应用检验