To address the issue of weakly discernible wave heads in the reflection waves of faults in the fully parallel auto transformer (AT) traction network, compounded by difficult localization caused by refraction and reflection complexity of fault traveling waves introduced by the parallel structure of the lines, a single-ended localization method for fault traveling waves in the traction network is proposed based on variational mode decomposition (VMD) and enhanced energy operator. Firstly, the transmission characteristics of fault traveling waves in the traction network are analyzed, with a focus on the impact of the fully parallel structure on the refraction and reflection of fault traveling waves. Features of different fault types and power flows are determined, facilitating the extraction of fault features. The process of wave head identification is transformed into the extraction of abrupt energy changes. Then, VMD is used for denoising and extracting the genuine voltage traveling wave component. In view of the challenging calibration of weak wave heads in the second reflection wave, a sliding time window (STW) combined with the symmetrical differencing energy operator (SDEO) is employed to construct the second instantaneous energy spectrum of the fault signal, yielding satisfactory results. Simulation outcomes demonstrate the method’s robust resistance to transient resistances, its ability to reflect variations in electromagnetic energy under different operating conditions in the fault traction network, and its high precision in fault localization.