Aiming　at　achieving　composite　materials　damage　identification　of　Structure　Health　Monitoring　(SHM),　an　preprocessing　algorithm　of　Redundant　Second　Generation　Wavelet　(RSGW)　considering　neighboring　coefficients　is　introduced,　which　can　denoises　Lamb　wave　signals　from　noise　flooding　and　structure　complexity　conditions　for　damage　features　extraction　and　further　calculations.　The　practicability　is　validated　through　experiment　operated　in　composite　damage　monitoring　system.