Condition monitoring of electrical equipment using thermal image processing

Abstract

Excessive temperature rise leads to majority of failures in electrical equipment. Therefore, the thermal monitoring always plays a significant role for identifying incipient faults. Now a days cost effective, reliable and non contact type infrared thermographic inspection system is being widely utilized for monitoring and fault diagnosis. In this paper, thermal images of some electrical equipment have been taken and converted to HSI color model for further processing. Unlike other techniques, instead of gray scale images, in the proposed method hue region has been taken into consideration. Then different gradient based edge detection like Prewitt, Roberts and Sobel are used to identify hot region of the thermal image. Based on some image metrics like Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR), the best method is chosen for processing of the thermal image. Lastly, Clustering based Otsu image segmentation method is introduced taking hue region. These methods are tested on 27 numbers of standard thermal images. The proposed technique gives better segmentation results for all images than the standard grey scale approaches. Hence the proposed model may produce better features for thermal monitoring of electrical equipment in future.