Accurate detection of fault traveling wave signals is an important factor to ensure accurate and reliable fault location results. Aiming at the problem that Wigner Ville distribution (WVD) is prone to produce cross terms when detecting fault traveling waves, a fault traveling wave detection method based on improved variational mode decomposition (VMD) and WVD is proposed. Firstly, the selection of VMD parameters is determined by waveform similarity, and the improved VMD is used to adaptively decompose the line mode traveling wave signal, so as to extract signals of different frequency bands of fault traveling wave while retaining important fault information. Then, the original WVD signal is obtained by the calculation and superimposition of each component signal of the WVD, thus solving the influence of cross terms due to WVD. Finally, the energy spectral density distribution of VMD-WVD is obtained, and the good time-frequency aggregation of WVD is used to detect the wave head position from 150 kHz to 170 kHz frequency band of the fault traveling wave, so as to achieve accurate detection of the fault traveling wave head. Simulation results show that the proposed method is faster and more accurate than the wavelet and HHT fault traveling wave detection methods.