Machine Learning-Based Prediction of Breakdown Voltage in High-Voltage Transmission Lines Under Ambient Conditions

Reliability and safety of high-voltage transmission lines are essential for stable and continuous operation of a power system. Environmental factors such as pressure, temperature, surface contamination, humidity, etc., significantly affect the dielectric strength of air, often causing unpredictable voltage breakdowns. This research presents a novel machine learning-based predictive framework that integrates Paschen’s Law with simulated and empirical data to estimate the breakdown voltage $\left(V_{bk}\right)$ of transmission lines in various environmental conditions. The main contribution is to demonstrate that data-driven prediction of breakdown voltage $\left(V_{bk}\right)$ using a hybrid machine learning model is in agreement with physical discharge theory. The model achieved root mean square error (RMSE) of 5.2% and mean absolute error (MAE) of 3.5% when validated against field data. Despite the randomness of avalanche breakdown, model predictions strongly match experimental measurements. This approach enables early detection of insulation stress, real-time monitoring, and optimises maintenance scheduling to reduce outages, costs, and safety risks. Its robustness is confirmed experimentally. Overall, this work advances the prediction of avalanche breakdown behaviour using machine learning.