基于支持向量机回归的区域洪水频率分析

区域洪水频率分析是解决无资料地区设计洪水计算的重要方法之一。在过去的研究中,区域洪水频率分析主要基于经验方法以及传统的指标洪水法。随着数学理论的完善和人工智能算法的提出,越来越多的非线性模型逐渐应用于解决区域洪水频率分析的问题。在该背景下,提出了一种基于支持向量机回归(Support Vector Regression,SVR)的方法,并将其应用于湘江流域区域洪水频率分析,将分析结果与传统回归手段的计算结果进行对比分析。结果表明:与传统指标洪水法相比,SVR算法能基于较小的样本量对无资料地区设计洪水作出更加稳健的估计,为无资料地区设计洪水的估算提供了一种新思路。

Analysis of regional flood frequency based on support vector regression method

Regional Flood Frequency Analysis (RFFA) is an important method to estimate the design flood of ungauged site. The empirical method and index flood method are commonly used in past RFFA study. With the development of applied mathematics and artificial intelligence technology, more and more nonlinear models are gradually applied in this domain. Given the condition, a method based on support vector machine regression (SVR) is proposed to analyze the regional flood frequency in the Xiangjiang River basin, and the calculated results are compared with those of the traditional regression method. The results show that compared with the traditional index flood method, the SVR algorithm could make a more robust estimation of the design flood in the ungauged area based on a smaller sample size, which provides a new way to estimate the design flood in the ungauged area.