AI-Based Data Fusion of IEC Methods With Dashboard Analytics for Hydrophobicity Evaluation of High-Voltage Insulators
Open Access
- 9 April 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 56498-56515
Abstract
Reliable assessment of insulator surface hydrophobicity is critical for the safety and continuity of power transmission and distribution networks and electrified transportation. While IEC TS 62073 prescribes three established methods—Contact angle(CA), Surface tension(SF), and Spray method(SP) —their deployment is hampered by operator subjectivity, environmental sensitivity, and the lack of an integrated interpretation. We present an AI-assisted, IEC-faithful framework that unifies contact angle, surface tension, and spray method into a single decision pipeline and exposes actionable outputs via a maintenance-oriented dashboard. Methodologically, Contact Angle is automated using Linear Regression Intersection for robust left/right contact estimation; Surface Tension is automated via cotton-tip last-frame and stationery-frame detection to time droplet reformation without manual frame picking; and Spray method leverages Zoning Feature Extraction with k-nearest neighbors, optionally complemented by an SVD–Mahalanobis classifier for borderline cases. Experiments on porcelain type 52-1 insulators under control conditions (30 °C, RH > 70%) comprise 90 droplet sequences for Contact angle, Surface tension and 2,800 Spray method images. As a result, applying these algorithms across the three methods yields interpretation accuracy exceeding 90%. Decision-level fusion improves consistency in intermediate classes while preserving IEC interpretability, and the dashboard provides transparent confidence and trend views for field engineers. In conclusion, the proposed algorithms increase the accuracy and reliability of insulator test interpretation and reduce ambiguity among visually similar features, enabling standards-compliant, reproducible assessments suitable for predictive-maintenance workflows in utilities and rail operations.Keywords
Funding Information
- Research Strengthening Project of the Faculty of Engineering, King Mongkut’s University of Technology Thonburi (KMUTT), Thailand
This publication has 25 references indexed in Scilit:
- HSNet: A Novel Edge-Preserving Hierarchical Separable Network for Video Shadow DetectionCircuits, Systems, and Signal Processing, 2025
- Collaborative spatial-temporal video salient object detection with cross attention transformerSignal Processing, 2024
- Hydrophobicity classification of composite insulators based on convolutional neural networksEngineering Applications of Artificial Intelligence, 2020
- Centroid calculation of the blastomere from 3D Z-Stack image data of a 2-cell mouse embryoBiomedical Signal Processing and Control, 2020
- Naive Bayesian classifier for hydrophobicity classification of overhead polymeric insulators using binary image features with ambient light compensationHigh Voltage, 2019
- An exact model based dynamic contact angle algorithmMeasurement, 2018
- Centroid tracking and velocity measurement of white blood cell in videoHealth Information Science and Systems, 2018
- Introduction to Video and Image Processing2012
- KNN Model-Based Approach in Classification2003
- Hydrophobicity estimation of HV polymeric insulating materials. Development of a digital image processing methodIEEE Transactions on Dielectrics and Electrical Insulation, 2001