A system to detect small amounts of oil leakage with oil visualization for transformers using fluorescence recognition

A system has been developed for detecting and visualizing small amounts of oil leakage from a transformer. The system uses a method of detecting fluorescence emitted from oil when it is irradiated by black light. Image processing using the relationship between saturation and intensity in a hue-saturation-intensity color space is used to recognize the leaked oil. According to the saturation-intensity relationship, the intensity of a surface without oil adhesion linearly increases as saturation increases. The oil-adhered area on the surface showed higher intensity. An algorithm was constructed for automatically setting a threshold value line to recognize the oil. On the other hand, in order to recognize the specular reflection light noise, images obtained after black light irradiation from two directions were compared. The developed system using the algorithm can detect and visualize oil leakage from a transformer at an amount of less than 1 mL.