【作者】 李抗彬；

【导师】 李智录；

【作者基本信息】 西安理工大学 ， 农业水土工程， 2007， 硕士

【摘要】 洪涝灾害在我国从古至今都是十分突出的,它给国家和人们的生活带来重大的损失,因此对于水库入库洪水的预报研究具有重要的意义。本文针对新疆下坂地水库流域——塔什库尔干河流域的洪水特点,分别采用五种不同的建模方法,建立流域冰雪融水期的洪水径流预报模型,以期所建立的冰雪融水径流预报模型能够为下坂地水库安全运行、科学管理及流域水资源的合理利用提供可靠的数据支持。本文主要的内容及成果如下:(1)对流域水文气象资料进行整理,为流域洪水特性分析和建立流域冰雪融水径流预报模型作必要的准备;(2)分析流域冰雪融水型洪水形成的机制并对流域历史资料进行分析;(3)采用SRM冰雪融水径流预报模型模拟预测流域的径流过程,采用无量纲拟合优度系数和径流体积差两个模型精度评价指标,对SRM模型应用效果进行评价,结果表明SRM模型在该流域具有较好的应用效果;(4)采用多元线性回归模型建立流域的冰雪融水径流预报模型,并对模型进行检验和评价,结果表明该模型具有一定的应用价值;(5)采用偏最小二乘回归方法建立流域的冰雪融水径流预报模型,并对模型进行检验和评价。结果表明模型的稳定性、可靠性、及模型的预测结果相对多元线性回归模型有一定的提高,但模型的解释性并没有提高;(6)采用有限记忆静态灰色模型建立流域的冰雪融水径流预报模型,模型检验和评价结果表明:该模型对流域洪水期的径流过程有一定的预测模拟能力,但是模型预测结果中存在的尖点问题,需要在模型应用过程中特别注意;(7)采用RBF神经网络模型建立流域的冰雪融水径流预报模型,并对模型的预测结果进行检验和评价,检验和评价结果表明RBF神经网络模型的能够对流域的径流机制充分地描述,具有很大的应用价值;(8)对流域所建立的几种冰雪融水径流预报模型进行综合评价,为流域洪水期的洪水预报建立一个合理洪水预报方案。更多还原

【Abstract】 Damages of floodwater are always severe in China in all ages, it brings significant loss to the country and people’s life, so forecasting of inflow hydrograph of reservoir are important. According to the floodwater characteristics of valley in Xinjiang Xiabandi reservoir----Ta Shi Ku Er Gan river valley, use five kinds different methods to establish the runoff forecasting model of the valley in the floodwater period, to expect the model that can be able to provide the reliable datum of the reservoir safety running, scientific management and water resource reasonable using. In the article, the contents and results are as follows:(1) Cleaning up the valley hydrology and weather datum, make preparation for analyzing the flood characteristics and establishing the floodwater forecasting models;(2) Analyses the mechanism of floodwater forms and historical datum in the valley;(3) The SRM ice-snow melting runoff forecasting model is used for simulating the valley floodwater runoff process, and the Nash-Sutcliffe coefficient and volume dispatch is adopted as the evaluation standard for the SRM model, the result indicate that the SRM model has the very good application prospect in the valley;(4) The multi-linear regression analysis method is used for establishing the runoff forecasting model of ice-snow melting in the valley, and the model is tested and evaluated, the result indicates that the model having certain application value;(5) Using the partial least square method establishes the runoff forecasting model of ice-snow melting in the valley, through testing and evaluating of the model, the forecasting result of the partial least square model has more improving than the multi-linear regression analysis method, but the explanatory of the partial least square model has not improving;(6) Limited memory static grey model is used for establishing the runoff forecasting model of ice-snow melting in the valley, the testing and evaluating result indicates that the model has the forecasting capacity of the runoff process in the valley, but the sharp point problem in the model’s forecasting result, needs to pay attention toin the process of model application especially;(7) The RBF nerve network model is adopted to simulating and forecasting the ice-snow melting runoff process in the valley, the model’s testing and evaluating result indicates that the RBF nerve network model can describe the mechanism of runoff sufficiently, and has very important application value in the valley;(8) The five kinds ice-snow melting runoff forecasting model is carried out an overall evaluation, then, establish a reasonable floodwater forecasting project for the valley in floodwater period.更多还原