Understanding Corona Discharges using Digital Imaging

Corona discharges are unavoidable events in high voltage transmission system due to the existence of concentrated high electric fields. Some of the repercussions include power loss, generation of audio noise, radio frequency interference etc. Corona discharges are also regarded as one of the degradation inducing factors in polymeric insulators. Therefore, detection and analysis of these discharges may greatly help in accessing the severity of the damage on the insulators. The presently available tools include the daytime UV cameras which are expensive and qualitative in nature. However, the commercially available digital single lens reflection (DSLR) camera is equipped to capture these faint discharges with longer exposure time (>1 s) under darker ambience and images may be processed to extract useful information pertinent to the discharge intensity. Exploring the possible use of DSLR camera for corona detection and analysis could emerge as one of the low-cost condition monitoring tools. In the present study, AC corona discharges with voltage levels from 6 kV to 12 kV are generated using a multiple needle corona electrode. The discharge images are captured with a varying exposure time of 20 s, 25 s and 30 s, depending on the over and under saturation of the images. A color thresholding-based image segmentation algorithm is used to extract only the discharge region from the obtained images. A parameter called mean pixel intensity (I avg ) is computed as a quantitative indicator of discharge level and its variation with respect to the applied voltage and the exposure time is studied. A matrix to visualize the sparsity pattern of the I avg is constructed. The matrix entries indicated that, as applied voltage increases the overlapping in the I avg values become more evident. This is attributed to the effect of over saturation in the image pixels. The study shows that images of different voltage level can result in same I avg for the different exposure time and this may be avoided by carefully selecting the camera parameters. © 2018 IEEE