Machine building Gearbox Fault Diagnosis Based on EEMD-SVD

Gearbox is an important mechanical device to transmit power. In order to ensure the normal operation of gearbox under the condition of top load, high efficiency and high precision, it’s necessary to extract fault feature information using signal processing method and to further analyze and research gearbox fault. In the paper an improved de-noising method based on de-noising of singular value decomposition (SVD) is proposed, which sets threshold on the basis of standard derivation of the difference between adjacent singular values, and simulation is made. Through further research, combined it with ensemble empirical mode decomposition (EEMD), a new de-noising method based on EEMD-SVD (ensemble empirical mode decomposition and singular value decomposition) is derived, which is proved to be an effective de-noising method through simulation experiment. EEMD-SVD method is applied to fault diagnosis of gearbox and good results are achieved.