DG-LSTM-SA: Hyperparameters for Energy Forecasting Models

Hyperparameters for the DG-LSTM-SA model, a Deep Gated Long Short-Term Memory network with Self-Attention designed for power generation and load demand forecasting. The model was evaluated on three real-world energy datasets (NEPOOL, Yichang, and Solar-Energy) and outperformed ten baseline models. The dataset was authored by Guoqiang Sun and last updated on June 3, 2026.

Use Cases

• Benchmarking forecasting models based on the described DG-LSTM-SA architecture.
• Tuning neural network hyperparameters for energy time-series prediction.
• Analyzing computational efficiency trade-offs in attention-based forecasting models.
• Reproducing ablation studies for grid dispatch and operational decision-making models.
• Comparing error reduction (e.g., MAE) against standard RNN variants like LSTM and GRU.

Strengths

• Model demonstrated a reduction in Mean Absolute Error by more than 75% compared to standard RNN variants.
• Dataset is openly licensed under CC-BY-4.0.
• Model achieved competitive accuracy with distinct advantages in computational efficiency and training speed relative to state-of-the-art attention-based models.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The 5.5 KB size suggests a very limited scope, likely containing only a small set of parameters.

Provenance

Source

figshare

Freshness

Last updated 2026-06-03 17:32:01; freshness should be verified.