In order to solve the problems of precocious convergence, poor fault tolerance and slow convergence speed of MVO algorithm in fault location of active distribution network, a fault location method based on SMVO is proposed. Firstly, adaptive elite and mutation strategies are used to improve the population quality and global development performance of MVO algorithm. Secondly, WEP and TDR parameter updating mechanism based on nonlinear curve change is designed to balance the convergence process of MVO algorithm in the early and late iterations. Finally, SMA algorithm is introduced to further improve the solving ability and solution quality of the improved MVO algorithm. In addition, the adaptive multi-DG switching function and high fault tolerance evaluation function are constructed, and a dynamic partition processing method suitable for active distribution network is introduced, which effectively reduces the dimension of the algorithm solution. The comparison experimental results of MVO, IMVO, SMA and SMVO algorithms in the simulation examples of single point, multi-point and information distortion faults of IEEE33-node active distribution network show that compared with the other three algorithms, the SMVO algorithm has higher local exploration ability and iterative stability, and the positioning accuracy reaches 100%. The average iteration time is only about 0.37s, and precocious convergence is not easy to occur, which can well meet the needs of online fault location of active distribution network