Assessment of Transmission Line Icing Level Based on IEEFO-XGBoost

In recent years, snow and ice-related disasters have become increasingly frequent, posing a significant threat to the operation of power transmission systems. The prediction of icing on transmission lines is one of the critical issues for ensuring the safe operation of power systems. The core of this challenge lies in the accurate prediction of ice thickness and its development trend through the integrated analysis of multiple data sources, such as meteorological data, thereby providing a scientific basis for decision-making in power sector protection. This paper proposes a machine learning-based risk assessment mechanism for icing on transmission lines, utilizing historical meteorological datasets from several cold regions. The improved Electric Eel Optimization (IEEFO) algorithm is combined with the Extreme Gradient Boosting (XGBoost) model to construct the IEEFOXGBoost model for icing level assessment. A comparison is made between the IEEFO-XGBoost model and other models, including Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Logistic Regression (LR), using standard performance evaluation metrics such as accuracy, precision, recall, and F1 score. Regarding recall, the IEEFO-XGBoost model demonstrates exceptional performance, achieving a recall rate of 0.9998, significantly outperforming the other models (XGBoost: 0.9565, RF: 0.9130, LR: 0.9542).