Fault Diagnosis for Fan Based on Self-Adaptive Boosting SVM Ensemble

A self-adaptive enhanced support vector machine ensemble method is proposed.The proposed novel method combined with auditory spectrum feature is applied to identify of fan fault.The identification experiments of field measurement data proved that the proposed method is effective for fan fault diagnosis.The classification accuracy of normal fans is about 99%,and the classification accuracy of the type of faulty fans of proposed SVM ensemble method is 1.88%~2.50% higher than single SVM classifier.