In　the　current　construction　of　smart　grids,　the　problem　of　bird　damage　pose　a　serious　threat　to　the　safe　operation　of　the　transmission　line.　To　this　end,　this　article　proposes　a　method　of　preventing　and　controlling　bird　harm　based　on　the　precision　testing　of　the　bird＇s　nest　and　the　rating　of　bird　harm.　First,　the　Mask2Former　model　is　used　for　image　segmentation.　This　model　can　accurately　identify　the　bird＇s　nest　in　the　complex　background　based　on　the　Transformer　architecture.　After　that,　the　lightweight　deep　convolutional　neural　network　was　planned　to　extract　the　visual　characteristics,　and　the　two-way　encoder　representation　method　was　used　to　analyze　the　text　characteristics　to　study　the　deep　fusion　scheme　of　the　image　and　text　information.　On　this　basis,　this　article　is　designed　with　the　bird　harm　level　evaluation　model　of　time　and　space　semantic　information.　This　model　intends　to　use　the　time　and　space　diagram　neural　network　to　predict　the　trend　of　the　shape　of　the　bird＇s　nest　form,　and　combine　the　random　forest　algorithm　to　evaluate　the　bird＇s　harm　level.　The　experimental　results　show　that　the　method　proposed　in　this　article　has　achieved　remarkable　results　in　the　division　of　the　bird＇s　nest,　with　an　average　accuracy　rate　of　99.37%.　Finally,　this　article　designs　the　visual　interactive　interface　of　bird　damage　prevention　and　control,　and　verifies　the　effectiveness　of　the　method　of　preventing　the　prevention　and　control　of　birds.　Through　this　study,　it　provides　strong　technical　support　for　the　prevention　and　control　of　the　transmission　line,　which　helps　ensure　the　stable　operation　of　the　power　grid.　©　2025　IEEE.