Orthogonalized variational mode decomposition

This paper introduces a modification to variational mode decomposition (VMD) by promoting mode orthogonality in the minimization problem. The main idea is to actively transmit and receive non-orthogonal signal components between the modes by imposing weak orthogonality conditions. The approach, combined with the proportional value of filter bandwidth, effectively prevents mode duplication and enhances robustness of the decomposition against over-segmentation. Experiments considering a broadband synthetic signal show the improved performance of this method in comparison to the standard Variational Mode Decomposition (VMD). The sensitivity of the method to different filter bandwidth, levels of noise, and its effectiveness in handling over-segmentation are also examined. • The Variational Mode Decomposition (VMD) method is sensitive to prescribed parameters. • Overlapping signal components and noise limit the robustness of VMD. • Minimizing coherence in the variational problem promotes orthogonality and robustness. • An adaptive filter parameter improves the method’s performance for broadband signals.