简介：Thedatausedintheprocessofknowledgediscoveryoftenincludesnoiseandincompleteinformation.Theboundariesofdifferentclassesofthesedataareblurandunobvious.Whenthesedataareclusteredorclassified,weoftengetthecoveringsinsteadofthepartitions,anditusuallymakesourinformationsysteminsecure.Inthispaper,optimalpartitioningofincompletedataisresearched.Firstly,therelationshipofsetcoverandsetpartitionisdiscussed,andthedistancebetweensetcoverandsetpartitionisdefined.Secondly,theoptimalpartitioningofgivencoverisresearchedbythecombingandpartingmethod,acquiringtheoptimalpartitionfromthreedifferentpartitionssetfamilyisdiscussed.Finally,thecorrespondingoptimalalgorithmisgiven.Therealwirelesssignalsofftencontainalotofnoise,andtherearemanyerrorsinboundarieswhenthesedataisclusteredbasedonthetradionalmethod.Inourexperimant,theproposedmethodimprovescorrectrategreatly,andtheexperimentalresultsdemonstratethemethod’svalidity.