In order to solve the difficulty of detecting the traveling wave head in a distribution network, a method of traveling wave signal detection based on Stochastic Resonance and Variational Mode Decomposition (SR-VMD) is proposed. The traveling wave signal is preprocessed by the variable scale SR of particle swarm optimization. This can effectively improve the signal-to-noise ratio of the output signal. The output signal is decomposed adaptively by a VMD algorithm and applied to fault location. Then the traveling wave head is calibrated by a Teager Energy Operator (TEO), and the fault distance is obtained by the distance formula. The simulation results show that the method can extract fault information from the noise background, and improve the accuracy of fault location, especially the accuracy of high resistance grounding and grounding near voltage zero crossing. This work is supported by Natural Science Guiding Program of Liaoning Province (No. 20180550438).