Abstract: Support Vector Machine is small sample method based on statistic learning theory. It is a new method to deal with the highly nonlinear classification and regression problems. It can better deal with the small sample, nonlinear and high dimension, avoiding the problem of local minima that the neural network can not solve. This paper briefly introduced the Support Vector Machine and the basic tenets of the infiltration flow monitoring data processing applications, described how to make use of Support Vector Machine to build dam seepage statistical model and forecasting. By calculating and analyzing on the continuous seepage monitoring data of Yunlong reservoir dam and comparing with RBF neural network forecasting result, the result showed that this method has higher precision and generalization ability than RBF neural network.

摘要： 支持向量机是基于统计学习理论的小样本学习方法, 是一种处理高度非线性分类回归等问题的新方法, 它能较好地解决小样本非线性高维数, 避免了神经网络无法解决的局部极小问题。本文简要介绍了支持向量机的基本原理及其在渗流监测数据处理中的应用,论述了如何利用支持向量机建立大坝渗流统计模型和预报。通过对云龙水库渗流监测连续观测数据的计算和分析, 并与RBF 神经网络预测结果进行比较, 证明支持向量回归机在渗流监测中比RBF 神经网络预测精度更高, 具有良好的泛化能力。