Probabilistic Dynamic Line Rating Forecasting with Line Graph Convolutional LSTM

• DOI: 10.48550/arxiv.2411.12963
• Publication Date: 2024-11-19
• Authors: Min‐Soo Kim; Vladimir Dvorkin; Jip Kim
• Keywords: Line (geometry)

Dynamic line rating (DLR) is a promising solution to increase the utilization of transmission lines by adjusting ratings based on real-time weather conditions. Accurate DLR forecast at scheduling stage thus necessary for system operators proactively optimize power flows, manage congestion, and reduce cost grid operations. However, remains challenging due uncertainty. To reliably predict DLRs, we propose new probabilistic forecasting model graph convolutional LSTM. Like standard LSTM networks, our accounts temporal correlations between DLRs across planning horizon. The graph-structured network additionally allows us leverage spatial features improve quality predictions. Simulation results synthetic Texas 123-bus demonstrate that proposed significantly outperforms baseline models regarding reliability sharpness while using fewest parameters.