Abstract
The inter-turn insulations of inverter-fed motor and some wind turbine generator usually suffer repetitive impulse voltage with high magnitude and steep edges. It may result in a significant reduction of life-time in case of the existence of inter-turn insulation defects. Based on the previous work with an established test system consisting of auxiliary impulse induction units and novel non-contact partial discharge (PD) sensor, this research continuously focuses on the signal processing technique and feature extraction method in inter-turn void defect diagnosis for inverter-fed traction motor winding. The singular values of characteristic matrix, which is constructed by wavelet packet decomposition of sensor’s response signal, have been proved sensitive to the PD-related information and qualified for determining the existence of inter-turn PD activity. In order to achieve the inter-turn PD feature extraction in detail, a time-domain signal processing method of PD pulses identification and segmentation has also been proposed. The statistical analysis on the inter-turn PD pulse signals which were picked out from the sensor responses in reduplicative measurements show that, the PD amplitude and time-lag are generally gamma distributed, while their per-unit values are beta distributed. In fact, the parameters of those statistical distributions are able to reflect the characteristic of inter-turn void defect under certain impulse voltage. Therefore, a method based on Kolmogorov-Smirnov test (K-S test) is proposed in order to diagnose the severity of the inter-turn void defect.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call