Based　on　empirical　mode　decomposition,　a　wavelet　threshold　denoising　method　is　presented　to　solve　the　signal　denoising　problem　of　low　signal　to　noise　ratio.　By　contrast　with　wavelet　threshold　denoising,　the　experimental　results　proved　that　this　method　is　very　efficient　and　stable　in　denoising　when　the　signal　to　noise　ratio　is　rather　low.　Furthermore,　a　new　way　of　studying　the　dynamic　signal　from　structure　under　environmental　pulsating　load　is　proposed.