A method for rotor vibration fault diagnosis was proposed based on wavelet energy spectrum entropy and support vector machine (SVM) by fusing their advantages.The typical rotor vibration faults were simulated on the rotor test-bed and vibration signals were collected at the same time.The SVM model and multi-classifier for rotor typical faults diagnosis were established through training of the samples obtained

and the unknown rotor faults were recognized consequently.It is proved by practical applications that the method proposed in this paper for diagnosing rotor vibration faults is workable and effective.