With　the　massive　participation　of　new　energy　in　the　operation　of　distribution　network　(DN),　the　uncertainties　arising　from　both　load　and　renewable　generation　have　brought　challenges　to　its　secure　and　reliable　operation.　In　this　paper,　a　model　of　day-ahead　and　intraday　optimal　scheduling　based　on　dynamic　partitioning　(DP)　of　DN　is　constructed　to　realize　the　refinement　of　operation　in　both　temporal　and　spatial　dimensions,　as　well　as　to　improve　the　reliability　of　system　operation.　For　the　strong　fluctuation　characteristics　of　the　source-load　in　DN,　the　DP　is　achieved　under　the　premise　that　the　partition　structure　in　each　period　is　tight　and　the　internal　power　supply　has　the　ability　to　balance　the　power　fluctuations.　In　order　to　improve　the　regional　autonomy　of　DN,　a　day-ahead　multi-objective　optimal　scheduling　model　is　constructed　based　on　the　DP　with　the　optimization　objectives　of　weak　inter-partition　power　coupling,　lowest　deficiency　rate　of　flexibility,　and　lowest　operating　cost.　And　a　scheme　for　articulation　is　proposed　to　ensure　the　effective　guidance　of　day-ahead　for　intraday.　Then　the　intraday　dispatch　model　is　constructed　by　shifting　the　focus　to　the　incorporate　characteristics　of　DP　and　quickly　smooth　the　net　load　fluctuations.　Finally,　the　performance　and　reliability　of　the　proposed　models　are　verified　via　case　studies.