With the advantages of low line cost and high efficiency, DC micro-grid has become the novel trend of power distribution in the future, but its multi-source and multi-transform easily lead to the existence of multiple forms of ripple generation sources in the system. The existence of a large number of ripples not only directly affects the power quality of DC micro-grid, but also is not convenient for accurate billing, so the accurate detection of ripples is important for the reliable power supply of DC micro-grid. In this paper, a ripple detection method based on modified variational modal decomposition (MVMD) is proposed, which can be used to accurately detect steady-state ripple and transient ripple. In the MVMD method, the best decomposition scale is determined based on the spectral feature correlation, and the VMD algorithm decomposes the DC detection signal into a series of intrinsic modal components (IMF), Hilbert transform (HT) is applied to obtain a series of parameters of the ripple components and the starting-ending time of the transient ripple signal. The MVMD method is compared with the empirical modal decomposition (EMD) algorithm, and it is proved that the proposed method has higher detection accuracy and noise robustness than the EMD algorithm, and is suitable for the monitoring of DC micro-grid ripple.