A fault location method for collector lines based on sparse information difference is proposed to address the issue of accuracy being easily affected by the number of measurement points and traveling wave speed during the fault location process of multi branch network structures in collector lines. The fault area is divided according to the collector line. First, this paper extracts the first intrinsic mode function signal from the initial traveling wave signal through variational mode decomposition, and further decomposes it into an S matrix through the Stockwell transform. Then, the modulus matrix Q is calculated through the matrix S, and the energy similarity between all modulus matrices is calculated to establish a regional decision matrix to identify the fault area. Finally, a virtual fault point is set within the determined fault area. Based on the initial arrival time of the traveling waves at the beginning and end of the fault area, the information difference between the measured traveling waves and the virtual fault point is calculated. Discrete-continuous particle swarm optimization is used to simultaneously correct the fault branch and fault location. The simulation results show that this method does not require traveling wave information from all branches of the collector line, significantly reduces the number of monitoring devices, eliminates the influence of wave speed, and achieves accurate fault localization of multi branch collector lines.