The　mechanism　of　Hematite　furnace　roasting　process　is　complex　and　operating　conditions　change　frequently　which　makes　it　difficult　to　get　the　set　value　of　controlled　variables　in　the　furnace　roasting　process　and　makes　it　tough　to　control　the　controlled　variable　within　the　target　range.　In　this　paper,　data　mining　technology　is　put　forward　to　solve　this　issue.　First,　clustering　method　is　used　to　deal　with　the　sample　data,　and　then　the　method　of　association　rules　in　data　mining　is　applied　to　obtain　association　rules　which　meet　the　conditions.　In　the　process　of　production,　the　correct　values　can　be　acquired　through　the　association　rule　table　which　provides　new　idea　for　the　optimization　settings　of　shaft　furnace　roasting　controlled　variable.　©　2013　IEEE.