A dagging‐based deep learning framework for transmission line flexibility assessment

Abstract Uncertainty in renewable energy generation, consumption, and electricity prices, as well transmission congestion, pose a number of problems modern power grids, necessitating stability on the supply, grid, demand sides. Grid‐side can be achieved by dynamic line rating (DLR) forecasting, which reliably predicts overall current carrying potential overhead lines. Long short‐term memory proved beneficiary this field, owing to its ability learn highly variable uncertain data. To empower network tackle non‐stationary nature meteorological parameters, novel machine learning (ML) architecture based Dagging technique is proposed tested data collected from 400 kV line. Simulation results corroborate that Dagging‐based stacked LSTM successfully handle issue outperform decomposition‐based technique, state‐of‐the‐art algorithm, for various forecasting horizons. The confirm generalizability models with an application DLR over without utilizing additional sensors communication networks. Moreover, model compared several ML architectures, including support vector machines (SVM), random forest (RF), multi‐layer perceptron (MLP) comprehensive benchmark study. introduced algorithm outperforms MLP 3.4%, RF 9.4%, SVM 6.7% terms average prediction accuracy.