Early Fault Diagnosis of Medium Voltage Covered Conductor Overhead Lines

Smart grids have emerged as a significant area of research due to rapid technological advancements and increased global focus on renewable energy. The reliable operation of overhead wires, a crucial component of the national electricity system, has a profound impact on both the national economy and people's livelihoods. This article presents a novel and specialized insulation fault diagnostic technique for covered conductor overhead lines using a lightweight gradient boosting decision tree (LightGBM). The proposed technique enables early detection of defects in covered conductor overhead lines by analyzing voltage signals collected from numerous sets of overhead lines under natural environmental conditions. The process involves signal processing, feature extraction, and algorithm analysis. Empirical results demonstrate that the developed model exhibits exceptional accuracy and stability, as measured by the Matthews Correlation Coefficient (MCC), surpassing other diagnostic models in terms of performance.