Abstract:

It is of great interest to extract motor slot harmonics for the purpose of studying motor efficiency, heat loss, torque ripple, as well as monitoring motor health condition and estimating motor speed. However, slot harmonics are typically of small magnitudes and of varying frequency in practice for motors operating at various conditions. In this paper, we propose a robust approach to extracting motor slot harmonics from the stator current when the motor is operating at varying speed conditions. First, a sparsity-driven method is proposed to estimate the frequency variation of slot harmonics due to varying operating speed. Then a singular value decomposition(SVD)-based method is used to extract the slot harmonics after compensating the corresponding frequency variation for each time window. In addition, the motor speed is also estimated according to the relationship between the slot harmonics and the motor speed. Experiments show that our method can extract slot harmonics from very noisy measurements and estimate motor speed with improved performance compared to the traditional method.