【作者】 赵雪花；

【导师】 黄强；

【作者基本信息】 西安理工大学 ， 水文学及水资源， 2005， 博士

【摘要】 河川径流作为水循环的重要环节,是水资源开发利用和科学管理重要的依据。然而河川径流的形成越来越多地受到人类活动和全球气候变化的影响,因此充分挖掘和识别河川径流的演变规律具有十分重要的理论价值和现实意义。 论文结合国家自然科学基金项目,以黄河上游兰州以上流域为研究对象,较为全面地分析了年径流量和各种影响因素之间的关系;提出了对河川径流序列进行挖掘与识别的思路和方法,从相似性、周期性和序列模式三方面对河川径流序列进行了挖掘。从而为河川径流演变规律挖掘提供理论基础。论文的主要研究内容和取得的成果有以下几个方面: (1)河川径流演变的影响因素相互作用、相互耦合,其作用机理十分复杂。受到当前科学技术发展程度的限制,单纯对某个或某几个影响因素进行分析往往具有一定的片面性,很难从整体上对其进行把握;单纯从定性的层面上讨论又很难满足实际应用的需要。本文针对以上问题对各种影响因素进行考察,分析各因素之间的影响与联系,探讨多种因素的综合效应,并对某些因素的影响进行了定量分析。 (2)首次将蚁群算法引入径流规律分析计算领域,并建立了基于蚁群算法的聚类分析模型,对多种径流影响因素进行综合聚类分析,通过相似性搜索对径流序列进行深入挖掘。 (3)运用极大熵谱分析、Hilben-Huang谱对径流时间序列进行了周期分析,指出了黄河上游径流演变可能存在的几个周期及其成因分析。Hilben-Huang变换是一种更多还原

【Abstract】 As a main link of water cycle, river runoff is the most important base for comprehensive development and utilization, scientific management and optimal operation. However, river runoff is affected by human activity and global climate change, so it is very important theoretical and practical sense for mining and identifying change law of the river runoff.In this paper, combined with the National Natural Science Foundation of China, the relation are analyzed between annual runoff and its factors, new methods are introduced for mining and identifying change law of river runoff, and similarity, periodicity and sequential patterns of runoff series are studied, which provide theoretical base for mining change law of river runoff. The analysis and calculation are conducted in combination with a real example, the watershed above the Lanzhou in the upper reaches of the Yellow River. The main research content and results are as follows:(1) Factors of river runoff change are interactivity and interdependency, and their mechanism of action is very complex. Limited by the present science-technology level, one or some factor is only analyzed, which is of one-sidedness. Moreover, qualitative discussion difficultly meets practical application need, so this paper analyses influenceand relation among factors, comprehensive effect of some factor, and quantitatively calculates some of factors.(2) Ant colony optimization (ACO) algorithm is firstly introduced the field of the runoff change analysis. Cluster analysis model is established based on ACO, by which some runoff factors are clustered. Therefore, runoff series is mined by means of similarity search.(3) Periodicity of runoff is calculated by maximum entropy spectrum analysis and Hilbert-Huang spectrum, the result shows some periodicities of runoff change and its cause in the upper reaches of the Yellow River. Hilbert-Huang transform is a new method for analyzing nonlinear and non-stationary data. This method is applied to mine and identify runoff series in this paper, by which runoff series is decomposed as various scale components and trend. These components are usually of clearer physical sense in comparison with classical method, so it provides a kind of effective way for analyzing runoff series.(4) This paper introduces a kind of model for forecasting runoff, life cycle model. Long-term trend of runoff change is forecast by it in the upper reaches of the Yellow River. This model considers that water resource is finite and runoff process is non-linear, and is a search for runoff forecasting.(5) Because runoff is affected by many factors, runoff process shows strong uncertainty and randomness. Forecasting model of river runoff is established based on grey theory. Moreover, three kinds of model is applied to forecast annual runoff in the upper reaches of the Yellow River, in which GM(1,1) is basic forecasting model, its forecasting result is not good, but may better reflect change trend of runoff. Grey topological model can realize waveform forecasting, however, forecasting precision is lower. Grey Markov chain model is based on the advantage of both GM(1,1) and Markov chain model, which may describe random fluctuating of runoff series, and improves the forecasting precision, so this is a kind of new and good search. It providesimportant reference value for real application.更多还原