Probability Prediction of Overhead Transmission Line DLR Based on EMD-XGBoost-MKDE

Dynamic line rating (DLR) holds significant importance in fully tapping transmission potential of overhead transmission line compared to static line rating (SLR). However, a single predictive model is insufficient to accurately capture the impact of meteorological elements such as wind speed and ambient temperature on the fluctuation and stochasticity of DLR. Therefore, the empirical mode decomposition algorithm is adopted to preprocess various meteorological data to reduce non-stationarity. Based on this, a point prediction model of DLR based on ensemble learning XGBoost is proposed. To account for the uncertainty of DLR, a mixed kernel density estimation algorithm is introduced to fit the probability distribution of DLR prediction errors, thereby achieving probabilistic prediction of DLR. Case studies demonstrate that the proposed method achieves superior predictive accuracy.