The　safety　of　prestressed　steel　structures　in　service　has　been　studied　widely.　However,　traditional　safety　assessment　methods　for　prestressed　steel　structures　involve　few　sample　points,　do　not　provide　accurate　predictions,　and　consume　substantial　human　and　material　resources.　The　digital　twin　technology　can　be　used　to　monitor　the　structural　behavior,　state,　and　activity　of　a　steel　structure　throughout　its　life　cycle,　which　is　equivalent　to　performing　a　safety　assessment　of　the　structure.　The　purpose　of　this　study　is　to　establish　a　digital　twin　multidimensional　model　of　prestressed　steel　structures.　Based　on　this　model,　the　support　vector　machine　and　prediction　model　are　trained　using　the　relevant　structural　history　data,　and　the　safety　risk　level　of　the　structure　is　then　predicted　based　on　the　measured　data.　Finally,　a　proportional　reduction　model　of　the　wheel-spoke　cable　truss　structure　is　used　to　verify　the　feasibility　of　the　proposed　method.　The　results　show　that　digital　twin　technology　can　achieve　real-time　monitoring　of　prestressed　steel　structures　in　use　and　can　provide　timely　predictions　of　the　safety　level.　This　represents　a　new　method　for　the　safety　risk　assessment　of　prestressed　steel　structures.