| Title |
Automatic Ultrasound Detection Exploiting Spectral Energy Analysis for Power Facility Diagnosis |
| Authors |
박준형(Junhyeong Pak) ; 심윤보(Yoon Bo Shim) ; 정상오(Sangoh Jeong) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.5.776 |
| Keywords |
Power Distribution Facility; Ultrasound Diagnosis; Spectrogram; Statistical Model; Ultrasound Abnormality |
| Abstract |
Deterioration such as erosion, corrosion, and crack in power facilities causes severe failure on power distribution system. In general, subjective listening assessment of audible sound from frequency-modulated ultrasonic signal of power facility is one of the most famous diagnosis techniques, and human errors lead to accuracy degradation of facility diagnosis. In this paper, we propose a novel approach related to automatic ultrasound detection based on spectrum analysis in order to minimize human error for power facility diagnosis. Specifically, the proposed method exploiting spectral energy obtained from spectrogram can automatically detect signal abnormality. Finally, we utilize the Mahalanobis distance to consider both stochastic and spectral analysis related to ultrasound abnormality. Experimental results show that the proposed method can analyze accurately the status of overhead power distribution system comparing to the conventional ultrasound diagnosis. |