Fault diagnosis for fan based on auditory spectrum feature of sound signal

In this paper, an auditory spectrum feature extraction method is proposed the psychophysics of hearing is simulated by auditory spectrum feature. Support vector machine (SVM) is used for the identification of fan fault. A fault diagnosis system is designed based on auditory principle and SVM, and is also applied to fault diagnosis for fan. A series of experiments were carried with the real data measured from a factory environment. The results show that the auditory spectmm feature is effective for fan fault diagnosis. The classification accuracy for normal fans is about 99.18%, and the classification accuracy of the types for faulty fans is above 91%.