The structure of RV reducer is complex,the vibration signal generated during the operation will be interfered by a lot of random noises due to the vibration excitation source. It will affect the real vibration signal be identified. The wavelet transform theory and the principle of signal recognition analysis and noise reduction analysis are introduced. The vibration signal of domestic SHPR-20 E RV reducer is analyzed based on wavelet transform theory,and signal recognition and noise reduction of vibration signal is analyzed. Through the test and analysis,it is shown that the wavelet transform has a good multi-resolution recognition property in time-frequency domain and can eliminate the noise effectively and extract the fault characteristic information of the signal. It provides a reference value for signal identification and analysis,fault diagnosis and extraction of RV reducer.