Condition recognition of induction machine under various loads

This paper offers a complete method to deal with the condition recognition for induction machines under different loads. In actual operation, the loads of induction machines vary within a certain range. When the signal is transformed into spectrum, the load variation causes band drifts and amplitude variation. The load factor influences the result of recognition seriously. This paper offers band adjustment and feature extraction to solve the recognition problems caused by loads and faults. Employing band adjustment, the method eliminates the phenomenon of band drift under different operation condition. Employing feature extraction, the method decreases the sensitivity of features to loads. Then, load influences on condition recognition for an induction machine will descend to a minimum. The method afforded in this paper is verified by experiment, and proves its feasibility with high accuracy.