Metro system plays a vital role in urban transportation, serving as lifeline for millions of commuters. However, the vulnerability of metro infrastructure to earthquakes underscores the necessity for robust recovery strategies. When an earthquake occurs, some inspections and maintenance works are required for affected stations to ensure the safe train operation. This study explores the optimal recovery sequence for affected stations, taking into account variations in maintenance team configurations over time. Based on graph theory, the topology structure of metro network is constructed, where the network average efficiency is selected as the metric for assessing network resilience performance. For each time-varying configuration of maintenance team, all combinations of recovery sequences are enumerated, and the corresponding network resilience performance is then calculated. Numerical experiments are carried out based on a simplified metro network. The computational results demonstrate the effectiveness of the proposed approach and indicate that the optimal recovery sequence is closely related to the node degrees of affected stations, network connectivity, and maintenance team configurations.