SavedAdd your journal

Journals / Balkan Journal of Electrical and Computer Engineering / 2026 / Cilt: 14

Short-Term Spinning Reserve Requirement Estimation Using TimeGAN-Based Synthetic Data Augmentation and a Hybrid LSTM–XGBoost Model

TimeGAN Tabanlı Sentetik Veri Artırımı ve Hibrit LSTM–XGBoost Modeli Kullanılarak Kısa Dönemli Döner Rezerv Gereksiniminin Tahmin Edilmesi

Yasin Sönmez(BATMAN ÜNİVERSİTESİ)0000-0001-9303-1735
Pages
101–108
Publication Date
2026-03-29
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The results confirm that the proposed TimeGAN–LSTM–XGBoost framework provides a robust, scalable, and highly accurate solution for short-term spinning reserve forecasting, with strong potential for real-world deployment in power system operation and energy markets.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Accurate short-term forecasting of spinning reserve requirements is essential for ensuring frequency stability, operational reliability, and economic efficiency in modern power systems.
📊
Key Findings & Data:This study proposes a novel hybrid forecasting framework that integrates synthetic data generation, deep learning, and machine learning to overcome these limitations.
💡
Conclusion & Impact:The results confirm that the proposed TimeGAN–LSTM–XGBoost framework provides a robust, scalable, and highly accurate solution for short-term spinning reserve forecasting, with strong potential for real-world deployment in power system operation and energy markets.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Döner Rezerv Tahmini · TimeGAN · LSTM

Abstract

Accurate short-term forecasting of spinning reserve requirements is essential for ensuring frequency stability, operational reliability, and economic efficiency in modern power systems. However, the increasing penetration of renewable energy resources and the limited availability of high-quality operational data make reliable forecasting a challenging task. This study proposes a novel hybrid forecasting framework that integrates synthetic data generation, deep learning, and machine learning to overcome these limitations. To the best of the author’s knowledge, this is the first study that integrates TimeGAN-based synthetic data generation with a hybrid LSTM–XGBoost model specifically for short-term spinning reserve forecasting. Given the limited availability of real-world spinning reserve datasets, this study employs a TimeGAN-based synthetic data generation approach trained on multivariate power system variables (load, renewable generation, and frequency) to construct a realistic and representative dataset for model development. First, TimeGAN is employed to generate realistic synthetic time-series data that preserve the temporal dynamics of load, renewable generation, frequency deviations, and spinning reserve patterns. This synthetic data is combined with real operational records to enhance the diversity and volume of the training set. Then, a Long Short-Term Memory (LSTM) model is used to capture long-range temporal dependencies, while XGBoost is applied to learn nonlinear and feature-driven relationships within the data. Finally, a hybrid fusion strategy based on both weighted blending and stacking regression combines the strengths of the two models. Experimental evaluations demonstrate that the hybrid model significantly outperforms individual models across all metrics. The stacking-based hybrid approach achieves the best performance with RMSE = 12.84 MW, MAPE = 4.21%, and R² = 0.965, outperforming LSTM and XGBoost by substantial margins. Additionally, the integration of TimeGAN reduces forecasting errors by up to 18% and improves generalization, highlighting its effectiveness in addressing data scarcity and privacy constraints. The results confirm that the proposed TimeGAN–LSTM–XGBoost framework provides a robust, scalable, and highly accurate solution for short-term spinning reserve forecasting, with strong potential for real-world deployment in power system operation and energy markets.

Özet

Döner rezerv gereksiniminin kısa dönemli olarak doğru biçimde tahmin edilmesi, modern güç sistemlerinde frekans kararlılığının sağlanması, işletme güvenilirliğinin korunması ve ekonomik verimliliğin artırılması açısından kritik bir öneme sahiptir. Bununla birlikte yenilenebilir enerji kaynaklarının elektrik üretimindeki payının giderek artması ve yüksek kaliteli operasyonel verilere erişimde yaşanan kısıtlar, güvenilir tahmin modellerinin geliştirilmesini zorlaştırmaktadır. Bu çalışma, söz konusu sınırlamaları aşmak amacıyla sentetik veri üretimi, derin öğrenme ve makine öğrenmesini bütünleştiren yeni bir hibrit tahmin çerçevesi önermektedir. İlk aşamada, yük talebi, yenilenebilir enerji üretimi, frekans sapmaları ve döner rezerv davranışlarının zaman dinamiklerini koruyan gerçekçi sentetik zaman serisi verileri üretmek için TimeGAN modeli kullanılmıştır. Üretilen bu sentetik veriler, eğitim veri setinin çeşitliliğini ve hacmini artırmak amacıyla gerçek operasyonel kayıtlarla birleştirilmiştir. Ardından uzun dönemli zamansal bağımlılıkları yakalayabilmek için Long Short-Term Memory (LSTM) modeli kullanılmış, verideki doğrusal olmayan ve özellik tabanlı ilişkileri öğrenmek amacıyla ise XGBoost algoritması uygulanmıştır. Son aşamada, iki modelin güçlü yönlerini bir araya getirmek amacıyla ağırlıklı birleştirme (weighted blending) ve stacking regresyon yöntemlerine dayalı bir hibrit füzyon stratejisi geliştirilmiştir. Deneysel değerlendirmeler, önerilen hibrit modelin tüm performans ölçütleri bakımından tekil modellere kıyasla belirgin şekilde daha üstün sonuçlar verdiğini göstermektedir. Stacking tabanlı hibrit yaklaşım en iyi performansı sağlayarak RMSE = 12.84 MW, MAPE = %4.21 ve R² = 0.965 değerlerine ulaşmıştır. Bu sonuçlar, LSTM ve XGBoost modellerine kıyasla önemli ölçüde daha yüksek doğruluk sağlamaktadır. Ayrıca TimeGAN entegrasyonu, tahmin hatalarını %18’e kadar azaltmış ve modelin genelleme yeteneğini artırmıştır. Bu durum, TimeGAN yaklaşımının veri yetersizliği ve veri gizliliği gibi sorunların aşılmasında etkili olduğunu ortaya koymaktadır. Elde edilen bulgular, önerilen TimeGAN–LSTM–XGBoost tabanlı hibrit çerçevenin kısa dönemli döner rezerv tahmini için sağlam, ölçeklenebilir ve yüksek doğruluk sağlayan bir çözüm sunduğunu ve güç sistemi işletimi ile enerji piyasalarında gerçek uygulamalar için önemli bir potansiyele sahip olduğunu göstermektedir.

🔖

Cite This Article

Sönmez, Y. (2026). Short-Term Spinning Reserve Requirement Estimation Using TimeGAN-Based Synthetic Data Augmentation and a Hybrid LSTM–XGBoost Model. Balkan Journal of Electrical and Computer Engineering, Cilt: 14, 101–108. https://doi.org/10.17694/bajece.1907986

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (27)

  1. Ortega-Vazquez, M. A., & Kirschen, D. S. (2007). Optimizing the spinning reserve requirements using a cost/benefit analysis. IEEE Transactions on Power Systems, 22(1), 24–33.
  2. Behrangrad, M., Sugihara, H., & Funaki, T. (2011). Effect of optimal spinning reserve requirement on system pollution emission considering reserve supplying demand response in the electricity market. Applied Energy, 88(7), 2548–2558.
  3. Tür, M. R., Ay, S., Shobole, A., & Wadi, M. (2019). Güç sistemlerinde ünite tahsisi için döner rezerv gereksinimi optimal değerinin kayıp parametrelerin dikkate alınarak hesaplanması. Gazi Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi, 34(3), 1171–1186. [4] Tür, M. R. (2018). Elektrik güç sistemlerindeki döner rezerv gereksiniminin ekonomik yönden incelenmesi (Doctoral dissertation). Yıldız Teknik Üniversitesi.
  4. Ardakani, F. F. A., Mozafari, S. B., & Soleymani, S. (2022). Scheduling energy and spinning reserve based on linear chance-constrained optimization for a wind integrated power system. Ain Shams Engineering Journal, 13, 101582.
  5. Saeed, R. M. M., et al. (2024). Maximizing solar share in robust system spinning reserve-constrained economic operation of hybrid power systems. Energies, 17(11), 2794.
  6. Asre, S., & Anwar, A. (2022). Synthetic energy data generation using time variant generative adversarial network. Electronics, 11(3), 355.
  7. Zhang, C., et al. (2018). Generative adversarial network for synthetic time series data generation in smart grids. In SmartGridComm 2018 (pp. 1–6). IEEE.
  8. Turowski, M., et al. (2024). Generating synthetic energy time series: A review. Renewable and Sustainable Energy Reviews, 206, 114842.
  9. Li, C., et al. (2019). Power load forecasting based on the combined model of LSTM and XGBoost. In International Conference on Pattern Recognition and Artificial Intelligence (pp. 46–51). ACM.
  10. Semmelmann, L., Henni, S., & Weinhardt, C. (2022). Load forecasting for energy communities: A novel LSTM–XGBoost hybrid model based on smart meter data. Energy Informatics, 5(S1), 24.
  11. Dakheel, F., & Cevik, M. (2025). Optimizing smart grid load forecasting via a hybrid LSTM–XGBoost framework: Enhancing accuracy, robustness, and energy management. Energies, 18(11), 2842.
  12. Gomez, W., et al. (2025). Short-term smart grid energy forecasting using a hybrid approach. Energy (early access).
  13. Wang, Z., & Hong, T. (2020). Generating realistic building electrical load profiles through the generative adversarial network (GAN). Energy and Buildings, 224, 110299.
  14. Acquaah, Y., et al. (2025). Realistic synthetic dataset generation for cyber-physical power systems. SN Applied Sciences, 7292.
  15. Yoon, J., Jarrett, D., & Van der Schaar, M. (2019). Time-series generative adversarial networks. In NeurIPS.
  16. Esteban, C., Hyland, S. L., & Rätsch, G. (2017). Real-valued (medical) time series generation with recurrent conditional GANs. arXiv preprint arXiv:1706.02633. [18] Song, L., Li, Y., & Lu, N. (2022). ProfileSR-GAN: A GAN based super-resolution method for generating high-resolution load profiles. IEEE Transactions on Smart Grid, 13(4), 3278–3289.
  17. Marino, D. L., Amarasinghe, K., & Manic, M. (2016). Building energy load forecasting using deep neural networks and XGBoost. In IEEE SMC.
  18. Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In KDD.
  19. Kong, W., Dong, Z. Y., Jia, Y., Hill, D. J., Xu, Y., & Zhang, Y. (2017). Short-term residential load forecasting based on LSTM recurrent neural network. IEEE Transactions on Smart Grid, 10(1), 841–851.
  20. Torres, J. F., et al. (2021). Deep learning for time series forecasting: A survey. Applied Soft Computing.
  21. O’Connor, C., Bahloul, M., Prestwich, S., & Visentin, A. (2025). A review of electricity price forecasting models in the day-ahead, intra-day, and balancing markets. Energies, 18(12), 3097.
  22. Yilmaz, B., & Korn, R. (2022). Synthetic demand data generation for individual electricity consumers: Generative adversarial networks (GANs). Energy and AI, 9, 100161.
  23. Fekri, M. N., Ghosh, A. M., & Grolinger, K. (2019). Generating energy data for machine learning with recurrent generative adversarial networks. Energies, 13(1), 130.
  24. Yarramsetty, C., Moger, T., Jena, D., & Rao, V. S. (2025). Advances in composite power system reliability assessment: Trends and machine learning role. IEEE Access.
  25. Shakeel, A., Chong, D., & Wang, J. (2023). District heating load forecasting with a hybrid model based on LightGBM and FB-prophet. Journal of Cleaner Production, 409, 137130.
  26. Catarino, A., Melo, R., Abreu, R., & Cruz, L. (2025). Evaluating privacy-utility tradeoffs in synthetic smart grid data. arXiv preprint arXiv:2506.11026.
  27. Phyo, P. P., & Byun, Y. C. (2021). Hybrid ensemble deep learning-based approach for time series energy prediction. Symmetry, 13(10), 1942.