DC load contains DC/DC converters and other power electronic devices, which is featured with significant nonlinear characteristics, resulting in a large number of ripple signals in DC output voltage and current signals. It is necessary to improve the accuracy of DC energy measurement by filtering and noise reduction. Aiming at the problem that the parameter setting of the existing filter de-noising method is not optimized and the effect of filter de-noising needs to be improved, this paper proposes an adaptive noise reduction method for metering signal of direct current electricity meter based on adaptive variational mode decomposition and wavelet threshold de-noising. The parameter optimization model of variational mode decomposition of DC voltage and current signal is established, and the mutual information analysis is combined to realize the adaptive differentiation between the effective mode component and the noise mode component of the original signal. On this basis, the optimization model of wavelet threshold de-noising parameters is established with the composite evaluation indices of signal-to-noise ratio, root-mean-square error, smoothness and correlation coefficient, and the optimal filtering of noise modal components is realized. The validity of the proposed method is verified by calculating and analyzing the measured data.