There are problems of slow convergence, ease of falling into local optima, poor fault tolerance and low population quality of existing intelligent optimization algorithms in solving active distribution network fault location problems. Thus this paper proposes an improved multi-verses optimization (IMVO) algorithm. First, an active distribution network model with fault tolerance is constructed, and the populations of the multiverse are discretized and coded according to the characteristics of the fault location problem. Second, an adaptive elite strategy is incorporated into the update of the multiverse population of the improved algorithm to ensure the population quality of the multiverse; the update mechanism of wormhole existence probability (WEP) and travel distance rate (TDR) based on nonlinear curve change is designed to improve the ability of searching the relative optimal universe in the front part of the algorithm and the accuracy of adjusting the optimal detection distance in the back part. Finally, the local search capability of the improved algorithm is enhanced by an adaptive mutation operation, thus improving the global search capability. The simulation results show that the improved multiverse optimization algorithm has significant global search capability in single-point, multi-point and information distortion fault location, and has a higher accuracy and convergence rate than other optimization algorithms in solving the distribution network fault location problem. This work is supported by the National Natural Science Foundation of China (No. 51179074).