Abstract：This paper proposes a noise reduction algorithm combining adaptive variational mode decomposition and KSVD dictionary learning. In this method, we fully retain the effective information in the monitoring sequence by denoising the sub-sequences after the decomposition of the monitoring sequence, and consider the features in the residual sequence. We take the deformation monitoring data of a dam as an example. The results show that the proposed method can effectively retain the effective information in the monitoring sequence, and is more suitable for dam deformation prediction under complex conditions than the traditional noise reduction method, and can further improve the generalization ability of the prediction model.
