Supervised Learning Approach to Infrared Image Based Condition Assessment of 33kV Lightning Surge Arresters
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This paper presents an online condition assessment method for 33kV Silicon rubber (SIR) housed ZnO surge arresters through infrared imaging. A total of 46 arrester samples including field-energized arresters, lab-tested arresters (field aged and a virgin arrester) were analyzed by infrared images and grouped into three "health" classes by laboratory measurements and modelling. Insulation resistance (IR) and leakage current (LC) measurements were taken for lab-tested arresters, whereas Finite Element Method (FEM) based modeling was used to assess field-energized samples. An infrared image consisted of 180 features and by Linear Discriminant Analysis (LDA), 180-dimensional thermal data were reduced to a 2D space with the three classification labels. Decision boundaries for classifying arresters were defined using Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN) classifiers with their accuracy evaluated for effective arrester condition assessment.
Fields of Study & Subjects
High voltage insulation and dielectric phenomena
Thermography and Photoacoustic Techniques
Power Transformer Diagnostics and Insulation
Publication details
- Journal / source
- IEEE International Conference on Industrial and Information Systems (ICIIS)
- Pages
- 506–510
- Publication date
- 16 January 2026
- Language
- EN