节点文献
水科学信息分析计算新方法及其应用
New Method and Its Application to Water Science Information Analysis and Calculation
【作者】 熊建秋;
【导师】 李祚泳;
【作者基本信息】 四川大学 , 水文学及水资源, 2006, 博士
【摘要】 水科学信息分析计算长期以来都是国内外研究的热点,一直处于积极探索和不断发展之中,特别是近一二十年,随着科学技术的进步,涌现了大量既有理论深度又有应用价值的研究成果,使这个领域的深入研究具有广阔的空间。 本论文依托973国家重点基础研究发展规划项目“长江中下游湖泊富营养化发展趋势预测方法研究”(NO.2002CB412301)和国家自然科学基金项目“基于子波和分形理论的水文尺度分析新途径”(NO.40271024),在总结吸收相关前人研究成果的基础上,基于大量实际水文水资源资料,运用现代智能科学的有关新理论和新技术,系统地研究和完善了部分水科学信息分析计算的新方法,提出了多种耦合预测模型,广泛适用于水科学和其它相关领域。此外,本文发现并明确指出了传统小波分解耦合预测方法的不足,并针对不足提出了信号倒置小波分解和信号滑动小波分解解决方案,对于科学合理地使用小波分析具有重要的指导意义和较高的实用价值。概括起来,本论文的主要研究内容和成果包括以下几个方面(各部分之间的联系请见第20页图1.1): (1) 全面系统地引入针对小样本数据且具有优良推广性能的支持向量机方法(SVM),将SVM初步应用于水电边坡稳定性预测、水流挟沙力预测和年用电量预测等实例,获得了较传统方法更好的效果;如何合理选择SVM的参数,目前仍缺乏有效的方法,这严重限制了SVM的实际应用,为此首先尝试引入了免疫进化算法(IEA)来优化SVM核函数参数,取得了一定的效果;考虑
【Abstract】 Water Science information analysis and calculation is continuously among development, and it is a hotspot, especially in recent 10-20 years. Along with the science and technology progress, many excellent research results come forth, making this territory a vast space worthy of study.Under the auspices of the National Key Project for Basic Research of China (No.2002CB412301) and National Natural Science Foundation of China(No.40271024), this thesis applies modern new theories and technique(such as Support Vector Machine, Particle Swarm Optimization, Artificial Neural Network, Wavelet Analysis, etc.) to systematically study the new methods in Water Science information analysis and calculation, and proposes various new hybrid models of prediction, which can be widely applied to the Water Science and other associated areas. Principal findings are concluded as follows.(1) Support Vector Machine(SVM) as a machine learning method is based on the solid theory foundation of Statistical Learning Theory, and focuses on the small samples. The theory of SVM and its characteristics were expatiated, and then proposed the application of SVM to the slope stability forecasting, sediment-carrying capacity forecasting and prediction of annual electricity consumption.