两种不同的SVM建模方法在大坝变形预测中的应用

摘要

Generally,there are two ways of modeling dam deformation monitoring data with support vector machine.First,the support vector machine model is constructed only with the dam deformation data as the input and output;second,the support vector machine model is constructed with the deformation of the dam impact factors such as temperature,water pressure as input,and the dam deformation data as output.There are few discussions about which modeling method is more outstanding.Two methods are used to model a dam in Fujian Province in this paper.Result shows that the second method not only can spend less time modeling and predicting,but also improve the prediction accuracy.%用支持向量机对大坝变形监测数据建模分析和预测一般有两种方法:一是仅用大坝的变形数据作为输入端和输出端,构建支持向量机模型;二是用温度、水压等大坝变形的影响因子作为输入端,大坝变形数据作为输出端,构建支持向量机模型.两种建模方法比较研究鲜有讨论,文中用这两种建模方法对福建省某一大坝进行建模预测.结果表明,第二种方法建模预测速度更快,预测精度更高.