论文部分内容阅读
针对局部均值分解LMD实现过程中存在的模式混淆现象,利用局部均值分解的原理,提出一种结合总体局部均值分解(ELMD)与最小二乘支持向量机(LSSVM)方法的多尺度大坝变形预测模型.利用ELMD方法对大坝变形序列进行分解,得到其PF分量,利用最小二乘支持向量机进行外推预测,再把各PF分量的预测结果进行叠加重构,得到大坝变形预测值.通过实例验证分析,比较多元回归分析、LSSVM和ELMD-LSSVM三种模型在大坝变形监测数据处理中的拟合和预测结果.研究结果表明:ELMD-LSSVM方法能够减弱模态混叠现象的影响,充分发掘数据本身所蕴含的物理机制和物理规律,为大坝变形多尺度预测分析奠定较好的基础.
Aiming at the pattern obfuscation existing in the process of local mean decomposition (LMD) implementation, this paper proposes a multi-scale dam deformation prediction method based on the principle of local averaging decomposition (LMMVM) combined with local local average decomposition (ELMD) and least square support vector machine Model, the dam deformation sequence is decomposed by using ELMD method, the PF component is obtained, extrapolation is predicted by least square support vector machine, and then the prediction results of each PF component are superimposed and reconstructed to obtain the dam deformation prediction value. The case validation and analysis were used to compare the results of fitting and forecasting of the dam deformation monitoring data with multiple regression analysis, LSSVM and ELMD-LSSVM.The results show that the ELMD-LSSVM method can reduce the influence of the modal aliasing phenomenon , Fully explore the physical mechanism and physical laws inherent in the data, and lay a good foundation for multi-scale prediction and analysis of dam deformation.