Focusing　on　the　problems　of　high　energy　consumption,　low　efficiency　and　poor　scalability　of　Practical　Byzantine　Fault　Tolerance　(PBFT)　consensus　algorithm　existed　in　consortium　blockchain,　this　paper　presents　a　group-based　optimized　Practical　Byzantine　Fault　Tolerance　(GPBFT)　consensus　algorithm,　which　is　a　multi-stage　algorithm.　First,　we　propose　a　comprehensive　reputation　evaluation　model　to　judge　the　credibility　of　a　node　from　two　aspects　of　transaction　behavior　and　consensus　performance.　Then,　the　nodes　with　higher　reputation　value　will　be　selected　to　enter　the　consensus　group,　and　the　other　nodes　with　block　packaging　ability　are　selected　as　candidate　nodes.　Finally,　we　optimize　the　PBFT　based　on　the　application　scenario　of　blockchain,　change　the　three　stages　to　two,　which　reduces　the　number　of　messages　delivered　by　50%　at　least.　The　experimental　results　demonstrate　that　GPBFT　gains　better　performance.　While　ensuring　the　security　and　reliability　of　the　system,　it　shows　good　scalability,　thus　it　can　be　used　in　large-scale　consortium　blockchain　system.　©　2020,　Springer　Nature　Singapore　Pte　Ltd.