There is a problem of deviation in fault location caused by uneven distribution of electrical energy in a DC distribution network. Focusing on the frequency band distribution characteristics of current mutation, a direct time reversal method is used to transform the distance problem into the response of the point current source field, and the translation invariance of a discrete dyadic wavelet is used at the same time. The time reversal fault transient traveling wave signal is reconstructed by the modulus maxima of the high-frequency component of the traveling wave signal in the form of pulse excitation. Its sharp change points are characterized accordingly, highlighting the characteristics of energy distribution. This is not limited by having little malfunction data. The applicability of the method in the main feeder and T-line is proved theoretically. The ±15 kV topology of a point-to-point MVDC distribution network model is built in PSCAD/EMTDC, and fault simulation and data acquisition verification are carried out considering the change of transition resistance and the influence of different fault types. The results show that the proposed method can realize the location in a short time window of 5 ms, and it has good anti transition resistance ability, which can reflect the change of electromagnetic energy of fault lines in different working conditions. This work is supported by the National Natural Science Foundation of China (No. 51877135).