Probabilistic day-ahead LMP forecasting for 8–12 PJM zones using Temporal Fusion Transformer and a frozen semantic-embedding fusion branch. Produces P10/P50/P90 hourly forecasts, attention and variable-importance visualizations, and fusion-gate diagnostics.
python deep-learning pytorch time-series-forecasting dvc probabilistic-forecasting mlflow energy-forecasting fastapi electricity-price-forecasting pytorch-lightning streamlit temporal-fusion-transformer ollama neuralforecast semantic-embeddings llm-fusion
-
Updated
May 28, 2026 - Python