摘要
变形是对大坝结构性态和安全状况最直接、可靠的反映,是大坝安全监测的重点项目之一。大坝变形具有较强的非线性特点,传统的预测方法有时精度不高。建立在统计学习理论和结构风险最小原理基础上的支持向量机算法能较好地解决小样本、非线性、高维数等问题。文章引入标准支持向量机的一种扩展——最小二乘支持向量机,参照传统逐步回归统计模型建模思想,建立了基于最小二乘支持向量机的大坝变形预测模型。通过紧水滩大坝变形实例计算,表明了该方法的可行性和优越性。
Deformation is the most direct and reliable reflection of structural state and safety situation in a dam, which is one of the key items in dam safety monitoring. Dam deformation is strongly non-linear and sometimes the effect of traditional forecasting methods is not very precise. Support vector machine based on statistical learning theory and structural risk minimization principle can deal with small samples, non-linear, high-dimensional problems more effectively. This paper introduces the LS-SVM which is an extension of support vector machines, and a dam deformation prediction model based on the LS-SVM was established according to the ideas of stepwise regression statistical model. Through calculation of Jinshuitan dam deformation, the results show this method is feasible and superior.
出处
《西北水电》
2011年第B09期53-56,共4页
Northwest Hydropower
基金
国家自然科学基金(51079114)
关键词
最小二乘支持向量机
变形预测
统计模型
least squares support vector machine
deformation prediction
statistical model
