摘要:To further enhance the early diagnose precision and efficiency for rotor crack under condition of strongly noise, AE signals is applied to extract fault feature of early rotor cracks with the advantages of strong anti-noise ability, clear fault feature. Based on the Pseudo Wigner-Ville Distribution (PWVD), the amplitude and frequency as feature vector are extracted from the AE signal in this paper, which are used to diagnose and predict the fault types of different rotor cracks depth by support vector machine (SVM). The experiment is done to verify the proposed method applied to fault diagnosis of cracked rotor.