In　order　to　determine　the　important　control　parameters　in　multi-gradient　drilling　and　optimize　drilling　efficiency,　a　dynamic　parameter　optimization　model　was　established.　In　the　improved　model,　the　maximum　drilling　depth　after　the　adjustment　of　the　control　parameters　for　a　single　drill　bit　trip　and　the　maximum　drilling　depth　under　the　same　well　structure　were　set　as　the　optimized　subjects.　The　position　of　cyclone　and　drilling　fluid　density　were　set　as　the　optimizing　parameters,　and　the　constraint　relationship　between　the　drilling　fluid　cyclic　equivalent　density　and　safe　density　window　as　well　as　the　drill　bit　life　expectancy　were　the　constraints.　The　optimization　model　was　solved　by　a　particle　swarm　optimization　algorithm.　The　results　of　case　study　show　that　there　are　optimal　position　of　the　cyclone　separator　and　density　of　the　drilling　fluid　after　each　adjustment　to　make　the　drilling　depth　of　each　single　trip　longest.　With　the　increase　of　the　number　of　the　adjustments,　the　optimal　position　of　the　cyclone　separator　from　the　drill　bit　and　the　optimal　density　of　the　drilling　fluid　are　gradually　increased,　but　the　maximum　drilling　depth　of　the　single　trip　is　gradually　reduced.　The　design　of　cyclone　separator　and　bit　life　expectancy　should　be　coordinated　to　avoid　waste　of　rising　and　drilling.　©　2019,　Periodical　Office　of　China　University　of　Petroleum.　All　right　reserved.