Characterization of cellulose bridging pattern in transformer oil using feature extraction technique

• Published in: 2022 IEEE International Conference on Power and Energy: Advancement in Power and Energy Systems towards Sustainable and Resilient Energy Supply, PECon 2022
• Main Author: Mustafa N.B.A.; Ramasamy I.D.; Nordin F.H.; Ali N.H.N.; Zainuddin H.; Daud M.M.
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2022
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85146418956&doi=10.1109%2fPECon54459.2022.9987705&partnerID=40&md5=5964ced54547e702f13c2b78418e08c9

Transformer oil or insulating oil is stable at high temperatures and has excellent electrical insulating properties. The most frequent problems occurred in transformer were related to the defects and weakness of the insulation systems. Many diagnostic methods have been introduced to provide the reliable assessment of insulating oil quality. Hence, in this work, image processing technique known as feature extraction is used to measure the cellulose bridging thickness in pre-bridging and bridging stage. These two stages were considered as early prediction before breakdown occurs. The cellulose bridging formation recorded in three selected types of transformer oils which include MIDEL 7131, PFAE and Gemini X mineral oil. The cellulose bridging images were captured from the bridging formation videos and 60 images were chosen from the extracted cellulose bridging images for each transformer oil. The 60 images were divided equally for pre-bridging stage and bridging stage. The captured images were then fed into feature extraction process to extract eight feature descriptors which include area, minor-axis length, major-axis length, orientation, contrast, correlation, energy and homogeneity. In our findings, the pixel values increase proportionally with the cellulose bridging thickness. Hence, distinct pattern of pre-bridging and bridging stages were illustrated in all feature descriptors. With this technique, an early prediction can be made to analyze the deterioration of transformer oil. © 2022 IEEE.