Small Amounts of Transformer Oil Leakage Fluorescence Detection Using Image Processing

Mental oil, as a major type of transformer insulating dielectric, plays a significant role in ensuring safe operation of transformers. However, oil leakage has become one of the most frequent defects because of ageing sealing elements. Thus, it is significant to propose efficient, intellectual and automatic detection methods of oil leakage for ensuring equipment severity. The characteristic that the fluorescence appears in ultraviolet light for mental oil is utilized in this paper. A detection method of oil leakage through automatic analysis and processing based on ultraviolet fluorescence images is proposed, which can identify transformer oil intellectually in the dark. First, the fluorescence images are analyzed through image processing methods. Based on morphological processing and Otsu threshold segmentation, the oil leakage was identified in YUV color space using background subtraction. Then, the filed ultraviolet fluorescent pictures are collected to testify the method. The field results indicate that the proposed method could effectively recognize the oil traces with fluorescence effect and their approximate areas, which lay the theoretical and experimental foundation for intelligent maintenance of transformer oil leakage.