Reverse　bond　percolation　is　the　inverse　process　of　bond　percolation,　which　examines　the　global　connectivity　of　a　network　after　adding　edges　with　uniform　probability.　Despite　its　universal　significance,　it　has　not　been　studied　as　extensively　as　bond　percolation.　This　paper　investigates　two　reverse　bond　percolation　scenarios:　in　the　first,　any　pair　of　unconnected　nodes　can　be　connected,　while　in　the　second,　only　specific　pairs　can　be　connected.　Two　typical　bond　percolation　models,　the　generating-function-based　model　and　the　message　passing　model,　are　reconstructed　for　each　scenario　to　calculate　analytical　values　of　the　spanning　cluster　fraction　and　percolation　threshold,　respectively.　These　analytic　models　are　then　applied　to　computer-generated　and　real-world　networks　to　assess　their　accuracy.　The　results　reveal　that　both　models　have　their　own　strengths.　While　the　message　passing　model　provides　more　precise　values,　it　requires　a　significant　amount　of　computational　effort　and　is　not　ideal　for　networks　with　a　large　number　of　existing　and　potential　edges.　The　generating-function-based　model　is　less　accurate　than　the　message　passing　model,　but　its　calculation　process　is　simple,　making　it　suitable　for　a　variety　of　real-world　networks.　These　two　reverse　percolation　analytic　models　may　complementarily　predict　percolation　properties　of　real-world　networks　after　the　edge-adding　process.&　COPY;　2023　Elsevier　B.V.　All　rights　reserved.