基于非平稳度量的灌溉用水量预测模型选择  

Selection of forecasting models for irrigation water consumption based on nonstationarity measure

作　　者：兰军[1] 谭秋衡[1] 董前进[1] 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北武汉430072

出　　处：《武汉大学学报（工学版）》2014年第6期721-725,共5页Engineering Journal of Wuhan University

基　　金：国家自然科学基金项目(编号:51190094;50909073)

摘　　要：非平稳性是水资源系统复杂性的一个重要方面,也是业界关注的理论难点之一.以稳定集合的概念为出发点,基于概率论中的中心极限定理及大数定律,探讨了可用于水文时间序列处理的稳定集合判别准则,并在仿真实验的基础上给出了选参经验公式以减小参数选择的主观性.在稳定信息结构的基础上给出了时间序列的非平稳度量指标并将其应用到模型选择问题中.借鉴AIC信息准则等方法,非平稳度量指标的提出为模型选择问题提供了一种新的准则:残差序列非平稳性度量最小化方法.利用该方法探讨农业灌溉用水量预测模型的优选问题,以长江支流举水流域的灌溉用水量预测为实例进行说明,并讨论了方法的有效性.Nonstationarity is an important aspect of the complexity of water resources system; it is also one of the most difficult theoretical problems that the related industry has focused on.The paper discusses the criterion for checking if a set is stable or not which can be used to process the hydrological time series based on the concept of stable set,the central limit theorem and the law of large numbers in the probability theory; and the empirical formula for selecting parameters is proposed based on simulation experiments.A nonstationarity measure index is introduced based on the concept of stable information structure.Furthermore,the index is used for selection of models and the new criterion for selection is proposed to minimize the nonstationarity measure of the residual sequence of a model.The authors discuss the selection of forecasting models for the irrigation water consumption of the Jushui river basin and the effectiveness of the method.

关 键 词：水文时间序列 非平稳性 稳定信息结构 灌溉用水量 模型优选 

分 类 号：P333[天文地球—水文科学]