To　locate　the　instant　and　location　of　abrupt　structural　damage　accurately,　this　paper　proposes　a　damage　identification　method　based　on　variational　mode　decomposition　(VMD)　and　the　autoregressive　integrated　moving　average　model　(ARIMA).　Initially,　the　method　accurately　selects　the　initial　central　frequencies　of　high-energy　modes　and　the　number　of　modes　to　be　decomposed　using　the　autoregressive　model　power　spectrum.　Then,　the　variational　mode　decomposition　method　is　applied　to　decompose　the　vibration　non-stationary　signal　into　several　stationary　signal　components,　followed　by　fitting　each　order　signal　component　using　the　ARIMA　model　to　obtain　model　residuals.　The　specific　time　of　damage　is　determined　by　analyzing　the　residuals　obtained　from　fitting　the　signal　components　with　ARIMA.　Finally,　the　structural　modal　shapes　are　obtained　using　principal　component　analysis　(PCA),　and　a　damage　index　that　integrates　frequency　and　modal　shape　is　constructed.　Combined　with　a　damage　threshold,　the　location　of　the　damage　is　pinpointed.　Furthermore,　the　method　is　analyzed　using　numerical　examples　of　a　ten-degree-of-freedom　structure　under　seismic　excitation　and　the　monitoring　acceleration　data　of　an　actual　steel　bridge.　The　results　demonstrate　that　the　proposed　method　can　accurately　locate　the　occurrence　time　and　location　of　structural　damage　under　both　stationary　and　non-stationary　excitations.　©　2025　Chinese　Vibration　Engineering　Society.　All　rights　reserved.