Bearing fault diagnosis method based on fuzzy support vector machine

The invention discloses a bearing fault diagnosis method based on a fuzzy support vector machine. Three faults, namely an inner ring single-point fault, an outer ring single-point fault and a ball single-point fault, are taken for example, and fault feature extraction, including demodulation on a time domain feature parameter and a vibration signal by Hilber transformation, is carried out by combining normal operation of a bearing; a demodulated signal is subjected to spectral analysis so as to find a frequency domain fault feature frequency. These feature parameters form a training sample and a test sample; a fuzzy membership degree is added into the training sample by a fuzzy C means clustering algorithm, and the fault judgment is carried out by a support vector machine multi-classification method. A fault diagnosis example analysis part displays the bearing fault diagnosis correctness of a constructed FSVM (fuzzy support vector machine) model, so that high classification performance and high noise resistance are achieved, and a theoretical method is provided in order to avoid accidents, economical loss and the like which are caused by the bearing fault; the bearing fault diagnosis method has important reference value.