In　rotating　machinery　equipment,　bearing　is　one　of　the　most　common　parts.　Because　of　the　complex　working　conditions,　the　bearing　system　is　subject　to　get　failure.　The　running　state　of　bearing　system,　which　is　normal　or　not,　will　directly　affect　the　safety　of　the　production　line,　or　even　cause　some　accidents.　Therefore,　the　technology　of　fault　diagnosis　of　rolling　bearing　has　important　theoretical　value　and　practical　significance　in　production　safety.　In　the　light　of　the　vibration　data　of　rolling　bearing,　including　the　normal　operation　of　rolling　bearing,　the　single　point　fault　of　the　inner　ring,　the　single　point　fault　of　the　outer　ring　and　the　single　point　fault　of　the　ball,　those　four　cases,　time,　envelope　and　frequency　analysis　were　performed　to　extract　fault　features.　Considering　the　interference　of　noise　and　outliers,　support　vector　machine　(SVM)　theory　combined　with　the　fuzzy　c-means　(FCM)　clustering　algorithm　was　used　to　establish　the　fuzzy　support　vector　machine　(FSVM)　model.　Train　the　samples,　using　the　founded　model　of　FSVM,　and　then　the　test　and　identification　of　bearing　fault　would　be　obtained.　©　2015　IEEE.