改进的LMD模型在大坝变形监测数据特征量分离中的运用

Application of improved LMD model in characteristic quantity separation of dam deformation monitoring data

运用改进的局部均值分解(LMD)模型对大坝变形监测资料物理特征量进行分离,通过分离出的物理特征分量来分析大坝变形效应量的影响因素及程度。与统计回归模型相比,LMD模型不依赖事先确定影响因子的数学表达式,便可以合理地分离出大坝变形效应量的物理特征分量,自适应性较强,而且在一定程度上比统计回归模型分解结果更优。

The improved local mean decomposition(LMD) model is used to separate the physical characteristics of dam deformation monitoring data. The influence factors and degree of dam deformation effect are analyzed by the separated physical characteristics. Compared with the widely accepted statistical regression model, the results show that the LMD model can reasonably separate the physical characteristic components of dam deformation effect without relying on the mathematical expression in which influencing factors are determined beforehand. It is more adaptive and, to a certain degree, better than the decomposition result of statistical regression model.