In　the　identification　process　of　SSI-DATA　method,　the　stabilization　diagram　is　generally　used　to　determine　the　order　of　the　system.　However,　in　the　modal　parameter　identification　of　spatial　grid　structure,　there　are　few　stable　points　that　meet　the　requirements,　and　the　stability　axis　is　not　obvious.　So,　the　location　of　the　modal　stability　point　cannot　be　determined.　Furthermore,　the　QR　decomposition　of　the　Hankel　matrix　is　computationally　inefficient　and　difficult　to　use　for　online　identification　when　analyzing　the　spatial　structure.　In　order　to　solve　the　above　problems,　this　paper　proposes　a　time-saving　random　subspace　method　of　spatial　grid　structure　based　on　an　improved　stabilization　diagram.　On　the　basis　of　the　traditional　random　subspace　method,　the　projection　of　the　Hankel　matrix　is　transformed　into　the　projection　transformation　of　its　R　matrix,　and　then　the　subsequent　calculation　is　carried　out.　Then,　the　frequency　search　domain　is　set　and　the　auxiliary　stability　axis　is　made.　According　to　the　spectrum　peak,　stabilization　diagram　and　the　auxiliary　stability　axis,　the　position　of　the　stability　axis　is　comprehensively　judged.　The　modal　parameters　of　the　structure　can　be　identified　more　accurately,　which　improves　the　calculation　efficiency　and　avoids　modal　omissions　or　misjudgments　to　a　certain　extent.　Using　a　spatial　structure　model　(upper　chord　18　x　12m,　lower　chord　16　x　10m,　grid　2　x　1.5m),　the　advantages　of　this　method　were　compared　and　verified.　The　results　indicate　that　the　calculation　time　of　the　traditional　SSI-DATA　method　is　29.077s,　while　the　improved　calculation　time　is　3.752s,　resulting　in　an　87.1%　improvement　in　calculation　efficiency,　while　also　preventing　mode　omission　and　misjudgment.　By　further　applying　this　method　to　vibration　monitoring　of　spatial　grid　structure　hangars,　its　applicability,　higher　computational　efficiency,　and　precision　have　been　proven.