Choosing the Right AI Model for Power Forecasting in Micro Gas Turbines: A Comprehensive Performance Study

Öz

This study evaluates the power prediction performance of fourteen different Artificial Intelligence models (traditional Machine Learning and Deep Learning) using a dataset of 71,225 samples from a 3-kilowatt micro gas turbine. The GRU architecture achieved the highest accuracy (RMSE: 12.36; R²: 0.9997), while traditional models such as XGBoost offered competitive performance with significantly lower computational requirements (77.13 seconds versus 855.04 seconds training time). Our findings demonstrate that optimal model selection depends on specific operational requirements, with traditional ML models being preferable for real-time applications and DL architectures for scenarios requiring high accuracy.

Anahtar Kelimeler

• Micro gas turbines
• energy forecasting
• deep learning
• time series analysis

Mikro Gaz Türbinlerinde Güç Tahmini İçin Doğru Yapay Zekâ Modelinin Seçimi: Kapsamlı Bir Performans Çalışması

Öz

Bu araştırmada, 3 kilowatt gücündeki bir mikro gaz türbininden elde edilen 71.225 örneklik veri seti kullanılarak, on dört farklı Yapay Zeka modelinin (geleneksel Makine Öğrenmesi ve Derin Öğrenme) güç tahmin performansı değerlendirilmiştir. GRU mimarisi en yüksek doğruluğu (RMSE: 12,36; R2: 0,9997) elde ederken, XGBoost gibi geleneksel modeller önemli ölçüde düşük hesaplama gereksinimleri (77,13 saniyeye karşı 855,04 saniye eğitim süresi) ile rekabetçi performans sunmuştur. Sonuçlar, optimal model seçiminin belirli operasyonel gereksinimlere bağlı olduğunu ve gerçek zamanlı uygulamalar için geleneksel Makine Öğrenmesi modellerinin, yüksek doğruluk gerektiren senaryolar için ise Derin Öğrenme mimarilerinin tercih edilmesi gerektiğini göstermiştir.

Anahtar Kelimeler

• Miikro gaz türbinleri
• enerji tahmini
• derin öğrenme
• zaman serisi analizi

Kaynakça

1. Su W, Zeng F, Shi M, Xiao X, Sun T, Guo J. A data-driven multi-time scale coordinated economic dispatch model for flexible interconnection low-voltage distribution station areas. In: 2022 Power System and Green Energy Conference (PSGEC); 25-27 August 2022; Shanghai, China. New York, NY, USA: IEEE. pp. 859-867.
2. Massaoudi M, Abu-Rub H, Refaat SS, Chihi I, Oueslati FS. Deep learning in smart grid technology: A review of recent advancements and future prospects. IEEE Access 2021; 9: 54558-54578.
3. Ahmadi B, Ceylan O, Ozdemir A. Distributed energy resource allocation using multi-objective grasshopper optimization algorithm. Electr Power Syst Res 2021; 201: 107564.
4. Ghenai C, Husein LA, Al Nahlawi M, Hamid AK, Bettayeb M. Recent trends of digital twin technologies in the energy sector: A comprehensive review. Sustain Energy Technol Assess 2022; 54: 102837.
5. Beucler T, Pritchard M, Rasp S, Ott J, Baldi P, Gentine P, ve diğerleri. Enforcing analytic constraints in neural networks emulating physical systems. Phys Rev Lett 2021; 126(9): 098302.
6. Robinson H, Pawar S, Rasheed A, San O, ve diğerleri. Physics guided neural networks for modelling of non-linear dynamics. Neural Netw 2022; 154: 333-345.
7. Gopisetty S, Treffinger P. Generic combined heat and power (CHP) model for the concept phase of energy planning process. Energies 2016; 10(1): 11.
8. Forootan MM, Larki I, Zahedi R, Ahmadi A. Machine learning and deep learning in energy systems: A review. Sustainability 2022; 14(8): 4832.

9. Gaboitaolelwe J, Zungeru AM, Yahya A, Lebekwe CK, Vinod DN, Salau AO. Machine learning based solar photovoltaic power forecasting: A review and comparison. IEEE Access 2023; 11: 40820-40845.
10. Saffari M, Khodayar M. Spatiotemporal deep learning for power system applications: A survey. IEEE Access 2024.
11. Wang J, Zhu H, Zhang Y, Cheng F, Zhou C. A novel prediction model for wind power based on improved long short-term memory neural network. Energy 2023; 265: 126283.
12. Jiang T, Liu Y. A short-term wind power prediction approach based on ensemble empirical mode decomposition and improved long short-term memory. Comput Electr Eng 2023; 110: 108830.
13. Pasandideh M, da Silva JA, Gonzalez-Hernandez JL, Mehr AS, Chen Y, Wang Y, Ting DSK, Carriveau R, ve diğerleri. Predicting steam turbine power generation: A comparison of long short-term memory and willans line model. Energies 2024; 17(2): 352.
14. Huang S, Yan C, Qu Y. Deep learning model-transformer based wind power forecasting approach. Front Energy Res 2023; 10: 1055683.
15. Khan ZA, Hussain T, Baik SW. Dual stream network with attention mechanism for photovoltaic power forecasting. Appl Energy 2023; 338: 120916.
16. Tian C, Niu T, Wei W. Developing a wind power forecasting system based on deep learning with attention mechanism. Energy 2022; 257: 124750.
17. Dong H, Zhu J, Li S, Wu W, Zhu H, Fan J. Short-term residential household reactive power forecasting considering active power demand via deep transformer sequence-to-sequence networks. Appl Energy 2023; 329: 120281.
18. Olsson T, Ramentol E, Rahman M, Oostveen M, Kyprianidis K. A data-driven approach for predicting long-term degradation of a fleet of micro gas turbines. Energy AI 2021; 4: 100064.
19. Belov S, Nikolaev S, Uzhinsky I. Hybrid data-driven and physics-based modeling for gas turbine prescriptive analytics. Int J Turbomach Propuls Power 2020; 5(4): 29.
20. Aslanidou I, Rahman M, Zaccaria V, Kyprianidis KG. Micro gas turbines in the future smart energy system: Fleet monitoring, diagnostics, and system level requirements. Front Mech Eng 2021; 7: 676853.
21. Souhaila C, Mohamed M. Ensemble methods comparison to predict the power produced by photovoltaic panels. Procedia Comput Sci 2021; 191: 385-390.
22. Piotrowski P, Baczyński D, Kopyt M, Gulczyński T. Advanced ensemble methods using machine learning and deep learning for one-day-ahead forecasts of electric energy production in wind farms. Energies 2022; 15(4): 1252.
23. Akilandeswari A, Raja SP, Shonika S, Anuradha S. Enhancing solar power generation forecasting using advanced machine learning and ensemble methods. In: 2024 International Conference on Power, Energy, Control and Transmission Systems (ICPECTS); 8-9 October 2024; Chennai, India. New York, NY, USA: IEEE. pp. 1-6.
24. Li J, Jia L, Zhou C. Probability density function based adaptive ensemble learning with global convergence for wind power prediction. Energy 2024; 312: 133573.
25. Hu Z, Jagtap AD, Karniadakis GE, Kawaguchi K. When do extended physics-informed neural networks (XPINNs) improve generalization? arXiv preprint arXiv:2109.09444, 2021.
26. Kapoor T, Wang H, Núñez A, Dollevoet R. Transfer learning for improved generalizability in causal physics-informed neural networks for beam simulations. Eng Appl Artif Intell 2024; 133: 108085.
27. Hashmi MB, Mansouri M, Assadi M. Dynamic performance and control strategies of micro gas turbines: State-of-the-art review, methods, and technologies. Energy Convers Manage X 2023; 18: 100376.
28. Mubarak H, Stegen S, Bai F, Abdellatif A, Sanjari MJ. Enhancing interpretability in power management: A time-encoded household energy forecasting using hybrid deep learning model. Energy Convers Manage 2024; 315: 118795.
29. Chen Z, Xiao F, Guo F, Yan J. Interpretable machine learning for building energy management: A state-of-the-art review. Adv Appl Energy 2023; 9: 100123.
30. Machlev R, Heistrene L, Perl M, Levy KY, Belikov J, Mannor S, Levron Y, ve diğerleri. Explainable artificial intelligence (XAI) techniques for energy and power systems: Review, challenges and opportunities. Energy AI 2022; 9: 100169.
31. Li G, Chen L, Liu J, Fang X. Comparative study on deep transfer learning strategies for cross-system and cross-operation-condition building energy systems fault diagnosis. Energy 2023; 263: 125943.
32. Li G, Zhang X, Mikulski M, Chen X, Ma Z, Ahn J, ve diğerleri. An improved transfer learning strategy for short-term cross-building energy prediction using data incremental. Build Simul 2024; 17: 165-183.
33. Chen L, Liu J, Li G, Wang Z, Zhang G, Liu Y, Li Y, Cheng L, ve diğerleri. Fault diagnosis for cross-building energy systems based on transfer learning and model interpretation. J Build Eng 2024; 91: 109424.
34. Bielski P, Eismont A, Bach J, Leiser F, Kottonau D, Böhm K. Knowledge-guided learning of temporal dynamics and its application to gas turbines. In: Proceedings of the 15th ACM International Conference on Future and Sustainable Energy Systems; 4-7 June 2024; Singapore, Singapore. pp. 279-290.
35. Ermshaus A, Schäfer P, Leser U. Window size selection in unsupervised time series analytics: A review and benchmark. In: International Workshop on Advanced Analytics and Learning on Temporal Data; 19-23 September 2023; Grenoble, France. Springer. pp. 83-101.
36. Maharana K, Mondal S, Nemade B. A review: Data pre-processing and data augmentation techniques. Glob Transit Proc 2022; 3(1): 91-99.
37. Probst P, Wright MN, Boulesteix AL. Hyperparameters and tuning strategies for random forest. WIREs Data Min Knowl Discov 2019; 9(3): e1301.
38. Nalluri M, Pentela M, Eluri NR. A scalable tree boosting system: XG boost. Int J Res Stud Sci Eng Technol 2020; 7(12): 36-51.
39. Ke G, Meng Q, Finley T, Wang T, Chen W, Ma W, Ye Q, Liu TY, ve diğerleri. Lightgbm: A highly efficient gradient boosting decision tree. Adv Neural Inf Process Syst 2017; 30.