Aiming　at　the　fault　diagnosis　of　rolling　bearing　in　the　case　of　complicated　background,　lifting　morphological　wavelet　is　used　to　denoise,　and　a　method　for　extracting　fault　features　is　represented　by　combining　lifting　morphological　wavelet　with　ensemble　empirical　mode　decomposition　(EEMD).　The　original　signal　is　denoised　firstly　by　max-lifting　morphological　wavelet　and　min-lifting　morphological　wavelet　filter　in　this　method,　then　fault　feature　information　is　extracted　by　obtained　intrinsic　mode　function　(IMF)　after　the　denoised　signal　is　decomposed　using　EEMD.　The　analysis　results　on　bearing　fault　vibration　test　signal　show　that　this　method　can　extract　fault　features　and　identify　fault　types　of　bearing　effectively.　©　2011　IEEE.