Automatic busbar detection in substation: Using directional Gaussian filter, gradient density, Hough transform and adaptive dynamic K-means clustering

As the power system developing towards the extra-high-voltage (EHV), ultra-high-voltage (UHV), and large capacity direction in China, online monitoring becomes much more vital for power equipment to ensure the safe operation of power system. The computer vision-based Equipment Map State Detection technique can capture the abnormal status of power device. The busbar, as a intermediate link at all levels of voltage distribution device, is vital in system safety and stability operation. In this paper, we propose an automatic busbar detection method in an image for substation using directional Gaussian filter to strengthen horizontal linear target in an image, gradient density to generate a busbar confidence map, Hough Transform to detect straight lines and adaptive dynamic K-means clustering to get closed lines for optimizing. Then, the busbar can be marked in image and it can be used for further checking whether there are foreign things hanging on the busbar. Experiments on some images taken in several substations demonstrate that our method is effective.