Machine Learning-Based Estimation of Transmission Line Tower Grounding Resistance Under Energized Conditions Using Residual Currents and Near-Surface Potentials

This paper proposes a novel methodology for estimating the low-frequency grounding resistance of transmission line towers under energized operating conditions by combining near-surface potential measurements with residual current monitoring. Unlike conventional Fall-of-Potential (FoP)-based methods, which require long distances and often demand de-energizing the installation, the proposed approach leverages the Residual Current Method (RCM) to operate reliably in energized and even urban environments. A high-fidelity three-dimensional finite-element model of a 525 kV self-supported tower was developed in COMSOL Multiphysics and used in conjunction with Latin Hypercube Sampling to generate a representative dataset encompassing various counterpoise degradation scenarios. Machine learning models, including Deep Neural Networks, Random Forest, XGBoost, and Support Vector Machine, were trained using locally monitored variables: five near-surface potentials and four counterpoise currents. Validation results demonstrated high predictive performance within the modeled domain, with an R2 value above 0.99 for grounding resistance estimation and an F1-macro score of up to 0.97 for multi-label classification of counterpoise health, even for a test set contaminated with noise up to 30dB. Feature importance analysis confirmed that the central near-surface potential is the most influential factor for regression, while tower-leg currents are the dominant factor for classification. Additionally, a reduced-feature assessment using only five variables yielded comparable performance, reinforcing the feasibility of large-scale implementation. From a practical perspective, the methodology enables real-time monitoring, facilitates anomaly detection such as counterpoise degradation or disconnection, and supports predictive maintenance strategies, contributing to improved reliability and operational safety in transmission systems.