With　the　rapid　development　of　smart　home　technologies,　various　smart　devices　have　entered　and　brought　convenience　to　people＇s　daily　life.　Meanwhile,　higher　demands　for　smart　home　services　have　gradually　emerged,　which　cannot　be　well　satisfied　by　using　traditional　service　provisioning　manners.　This　is　because　traditional　smart　home　control　systems　commonly　rely　on　manual　operations　and　fixed　rules,　which　cannot　satisfy　changeable　user　demands　and　may　seriously　degrade　the　user　experience.　Therefore,　it　is　necessary　to　capture　user　preferences　based　on　their　historical　behavior　data.　To　address　the　above　problems,　a　temporal　knowledge　graph　is　first　proposed　to　support　the　acquisition　of　user-perceived　environmental　data　and　user　behavior　data.　Next,　a　user-oriented　smart　home　service　prediction　model　is　designed　based　on　the　temporal　knowledge　graph,　which　can　predict　the　service　status　and　automatically　perform　the　corresponding　service　for　each　user.　Finally,　a　prototype　system　is　built　according　to　a　real-world　smart　home　environment.　The　experimental　results　show　that　the　proposed　method　can　provide　personalized　smart　home　services　and　well　satisfy　user　demands.　©　2022　ACM.