Skip to Main Content
The ageing of high and medium voltage power cable insulation is an increasing problem for the electric power industry, particularly under current economic constraints. To prevent insulation failure causing forced service outage, insulation condition monitoring techniques must be applied. Partial discharge (PD) measurement has been recognized as an effective diagnostic tool for insulation assessment. This is because most insulation failures start with PD phenomenon. However, on-site PD measurement signals are often affected by strong electromagnetic interference (EMI). Therefore, de-noising techniques have to be applied to recover the PD signal from noise to provide true information about the cable condition for asset management. This paper suggests using wavelet transform (WT) to reject on-site measurement noise. The existing wavelet de-noising methods are presented and their limitations led to the proposed new method. A novel WT de-noising threshold value selection method is introduced. Laboratory experiment results based on this novel method are presented. Its performance is compared with other WT de-noising methods. Results show this proposed method could greatly improve the measurement sensitivity and effectively remove on-site noise.