Urban　rail　transit　networks　are　essential　components　of　urban　transportation　systems,　but　they　are　vulnerable　to　disruptions　that　can　severely　affect　passenger　mobility　and　network　efficiency.　Traditional　methods　for　determining　restoration　sequences　often　rely　on　experiences　or　importance-based　approaches,　lacking　precision　in　identifying　critical　vulnerable　station　combinations　and　struggling　to　find　optimal　restoration　sequences　under　limited　budgets.　This　paper　introduces　a　three-level　model　framework　aimed　at　addressing　these　issues.　The　middle　and　lower　levels　jointly　identify　the　most　vulnerable　station　combinations,　while　the　upper　level　optimizes　the　restoration　sequence　by　taking　into　account　budget　constraints　and　changes　in　resilience　metric　throughout　the　restoring　period.　The　effectiveness　of　the　proposed　model　was　validated　using　four　subway　lines　in　Beijing,　China.　Results　demonstrate　that　the　model　can　effectively　identify　critical　vulnerable　station　combinations.　Additionally,　the　resilience-based　restoration　strategy　effectively　determines　the　optimal　recovery　plan　for　damaged　stations　under　limited　budgets,　outperforming　traditional　restoration　strategies　based　on　complex　networks　and　offering　strong　extensibility.