Solar PV Power Forecasting: Quantum, Hybrid, Deep Learning & ML Benchmark

A comprehensive benchmark for solar photovoltaic (PV) power forecasting, comparing quantum machine learning (QML), hybrid quantum-classical, and classical models on real-world PV data.

Designed for reproducibility and direct use in academic publications — IET conference paper template included.

---

Models Compared

Model	Type
Classical LSTM	Deep Learning
GRU	Deep Learning
Hybrid Quantum LSTM	Quantum-Classical
Quantum-Enhanced Model	Quantum ML
XGBoost	ML Baseline
CatBoost	ML Baseline
ARIMA	Statistical Baseline

---

Features

• Unified data pipeline for multi-year PV data (2022, 2023) with automatic cleaning and splitting
• Evaluation across MAE, RMSE, MBE, VAF, R², and MAPE on a held-out test set
• Publication-ready figures: training curves, error distributions, scatter plots, boxplots, bar charts
• IET conference LaTeX template for direct manuscript preparation
• All results, logs, and plots saved automatically

---

Data Setup

Place raw PV CSVs in:

data/raw/2022/   ← organized by month
data/raw/2023/   ← organized by month

The pipeline handles loading, cleaning, and stratified train/val/test splitting automatically.

---

Getting Started

1. Set up the environment

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Run training and evaluation

python src/training/Training_and_Evaluation_Script_for_Enhanced_QML_Models.py

By default, previously trained quantum and deep learning models are loaded from history/ to save time. To force a full retrain, set skip_trained_models=False in the script.

---

Outputs

Location	Contents
plots/	All publication-ready figures
history/	Model training logs and saved weights

Key figures include training_curves.png, feature_importance.png, and model_architecture.png.

---

Citation

If you use this benchmark in your research, please cite:

@software{mayan_sharma_2024_pvqml,
  author = {Mayan Sharma},
  title  = {Solar PV Power Forecasting: Quantum, Hybrid, Deep Learning, and ML Benchmark},
  year   = {2024},
  url    = {https://github.com/Mayan10/Enhanced-HQLSTM-Model}
}

---

Author

Mayan Sharma GitHub: @Mayan10

For questions or suggestions, open an issue on the repository.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.