论文部分内容阅读
主成分分析(PCA)能够有效地提取数据的特征信息,消除变量间的共线性,而将基于统计学习理论的支持向量机(SVM)用于数据建模具有显著的优点。本文将主成分分析应用到大坝变形影响因子的优化中,解决了由影响因子内部相关性而引入大量因子的问题,降低了输入维数,简化了输入结构。将简化后的数据作为SVM的输入因子,减少了SVM学习的时间,提高了拟合效率。试验结果表明该方法具有较高的预测精度和较强的泛化能力。
Principal component analysis (PCA) can effectively extract the feature information of the data and eliminate the collinearity between the variables. However, the support vector machine (SVM) based on statistical learning theory has significant advantages for data modeling. In this paper, principal component analysis (PCA) is applied to the optimization of the dam deformation factor, which solves the problem of introducing a large number of factors by the internal correlation of the influence factors, reduces the input dimension and simplifies the input structure. The simplified data is used as the input factor of SVM, which reduces the learning time of SVM and improves the fitting efficiency. Experimental results show that this method has high prediction accuracy and strong generalization ability.