As　the　energy　distribution　in　each　frequency　band　of　rolling　bearing　acoustic　emission　(AE)　signal　is　related　to　its　fault　type,　so　we　can　use　the　redundant　lifting　wavelet　packet　to　decompose　the　rolling　bearing　AE　signal　of　different　fault　into　different　frequency　band,　combine　energy　in　each　frequency　band　together　to　be　a　feature　vector　of　the　Support　Vector　Machines　(SVM),　then　being　applied　to　identify　the　fault　through　SVM.　This　paper　also　compared　the　redundant　lifting　wavelet　packet　and　Daubechies　wavelet　packet　as　well　as　the　SVM　and　neural　networks.　The　experimental　result　shows　that　for　the　fault　pattern　identification,　the　method　that　combines　redundant　lifting　wavelet　packet　decomposition　and　SVM　together　can　be　effective.