Many kinds of faults in high voltage power metering seriously affect the accuracy and reliability of the metering system. It is necessary to judge and identify the types of faults timely and accurately, eliminate the source of faults and restore the normal operation of the metering system. In this paper, the variational mode decomposition method is used to analyse the signals generated by the mixing of various frequency noises. The simulation and actual data validation show that the method can effectively extract different modes of complex signals, and then distinguish fault signals from normal signals, so as to improve the accuracy of fault identification.