A　new　fault　diagnosis　method　is　proposed　based　on　the　dual-tree　complex　wavelet　packet　transform　(DT-CWPT)　and　independent　component　analysis　(ICA),　aiming　at　separating　fault　information　from　the　compound　rolling　bearing　fault　signal.　First,　the　non-stationary　and　complex　signal　of　the　compound　fault　is　decomposed　into　several　different　frequency　band　components　through　dual-tree　complex　wavelet　packet　decomposition.　Then,　ICA　is　used　to　separate　the　mixed　signal　consisting　of　each　component　to　eliminate　the　frequency　aliasing　as　much　as　possible.　Finally,　independent　signal　components　separated　from　the　mixed　signal　are　processed　by　Hilbert　demodulation.　The　results　show　that　the　fault　feature　of　rolling　bearing　can　be　effectively　separated　and　extracted,　and　the　method＇s　feasibility　and　effectiveness　are　verified.　©,　2015,　Nanjing　University　of　Aeronautics　an　Astronautics.　All　right　reserved.