In　order　to　reduce　the　influence　of　uncertainty　of　renewable　distributed　generation　in　the　distribution　network　planning,　a　polymerization　method　of　output　curves　based　on　an　improved　K-means　clustering　algorithm　was　proposed.　The　typical　interval　output　scene　set　of　renewable　distributed　generation　(RDG)　was　constructed,　and　the　interval　power　flow　of　distribution　network　was　calculated　separately.　The　tolerance　of　the　uncertainty　for　RDG　was　defined.　The　multi-objective　optimization　model　was　established　based　on　the　NSGA-II　algorithm　aiming　at　minimum　system　network　loss,　minimum　voltage　offset　and　strongest　tolerance　of　uncertainty.　Finally,　the　IEEE　33　node　network,　the　actual　distribution　network　of　a　region　and　the　American　PG&E　69　node　network　were　used　as　examples　to　analyze　and　calculate　to　verify　the　correctness　and　effectiveness　of　the　proposed　model　and　method.　©　2020　Chin.　Soc.　for　Elec.　Eng.