The environmental noise is strong and complex during the operation of mining motor, and the early fault characteristics of the rolling bearing are easily overwhelmed. A singular value decomposition method of vibration signal was proposed with limited envelope spectral entropy (LESE) for the extraction of early fault characteristics of rolling bearings. According to the correspondence between the frequency and the singular value in the signal to be decomposed, the singular components corresponding to the same vibration signal component were accumulated as the output of a signal subcomponent; the entropy LESE was applied to solve the screening of the fault-sensitive signal components after the weak fault signal of the bearing was processed by SVD. Finally, the fault type of the rolling bearing was determined by envelope spectral analysis of the fault-sensitive signal components. The experimental results show that the proposed method can be used to realize the extraction of early fault characteristics of bearings, which is conducive to timely find bearing problems and avoid further deterioration of motor.