Zaman Serisi Analizi ile Enerji Tüketim Tahmininde İstatistiksel ve Makine Öğrenimi Yaklaşımları

Öz

Enerji üretimi, özellikle iklim değişikliğinin etkileriyle hızla büyüyen bir faaliyet alanı haline gelmiştir ve ülkeler arasında rekabet unsuru bile oluşturmuştur. Ancak, bu üretim çoğu zaman sabit veya sürekli olmamakta, hava koşulları veya bazı durumlarda fosil yakıt üretimi gibi dış faktörlere bağlı olarak değişiklik göstermektedir. Bu nedenle, enerji üretiminin verimliliğini optimize etmek ve yönetmek amacıyla tahmin edilmesi büyük önem taşımaktadır. Bu çalışmada, yenilenebilir enerji üretiminin zaman serisi tahminleri, ARIMA ve SARIMAX gibi istatistiksel modellerin yanı sıra LSTM ve Gauss Süreç Regresyonu (GPR) gibi makine öğrenimi modelleri kullanılarak gerçekleştirilmiştir. Kullanılan modeller, değerlendirme metriklerine, her modelin yaptığı tahminlere ve 72 adım boyunca yapılan öngörülere göre karşılaştırılmıştır. Uygulanan çeşitli karşılaştırma teknikleri sonucunda, en iyi performansı SARIMAX modeli göstermiş; bu model 0.000031 MSE, 0.0026 RMSE, 0.0015 MAE ve %99,98 R² değerlerine ulaşmıştır. Ayrıca, SARIMAX modeli verileri diğer modeller kadar etkili şekilde tahmin ederek neredeyse mükemmel öngörüler sağlamaktadır.

Anahtar Kelimeler

• ARIMA
• Enerji Üretimi
• Gauss Süreç
• LSTM
• Zaman Serisi

Statistical and Machine Learning Approaches for Energy Consumption Forecasting Using Time Series Analysis

Öz

Energy production is a rapidly growing activity, especially with the impacts of climate change. It has even become a competitive activity among countries. However, this production is not constant or continuous most of the time, as it depends on external factors such as weather conditions or, in some cases, fossil fuel production. Therefore, predicting energy production has become essential to optimize and manage its efficiency. In this study, a time series of renewable energy production is predicted using statistical models such as ARIMA and SARIMAX, as well as machine learning models such as LSTM and Gaussian Process Regression (GPR). These models are compared, based on evaluation metrics, on predictions made by each model, and on the forecasting over a period of 72 steps. After applying the various comparison techniques, the best-performing model is SARIMAX, with an MSE of 0.000031, an RMSE of 0.0026, an MAE of 0.0015 , and an R² of 99.98%. Furthermore, this model predicts the data as effectively as other models and provides near-perfect forecasting.

Anahtar Kelimeler

• ARIMA
• LSTM
• ARIMA
• Energy Production
• Gaussian Process
• LSTM
• Time Series

Kaynakça

1. [1] J. T. Hardy, Climate Change: Causes, Effects, and Solutions, John Wiley & Sons, 2003.
2. [2] U. Shahzad, "The need for renewable energy sources," Energy, vol. 2, no. 1, 2012.
3. [3] B. N. Stram, "Key challenges to expanding renewable energy," Energy Policy, vol. 96, pp. 728–734, 2016.
4. [4] C. Sweeney, R. J. Bessa, J. Browell, and P. Pinson, "The future of forecasting for renewable energy," Wiley Interdisciplinary Reviews: Energy and Environment, vol. 9, no. 2, p. e365, 2020.
5. [5] W. Sulistijanti and N. Khotimah, "Comparing time series predictions of COVID-19 deaths using SARIMAX, neural network, and XGBoost," Asian Journal of Engineering, Social and Health, vol. 3, no. 12, pp. 2751–2758, 2024.
6. [6] M. M. Rahman et al., "Prospective methodologies in hybrid renewable energy systems for energy prediction using artificial neural networks," Sustainability, vol. 13, no. 4, p. 2393, 2021.
7. [7] E. Connolly, "The suitability of SARIMAX time series and LSTM neural networks for predicting electricity consumption in Ireland," M.S. thesis, National College of Ireland, Dublin, 2021.
8. [8] F. R. Alharbi and D. Csala, "A seasonal autoregressive integrated moving average with exogenous factors (SARIMAX) forecasting model-based time series approach," Inventions, vol. 7, no. 4, p. 94, 2022.

9. [9] S. Siami-Namini, N. Tavakoli, and A. S. Namin, "A comparison of ARIMA and LSTM in forecasting time series," in Proc. 17th IEEE Int. Conf. Machine Learning and Applications (ICMLA), 2018, pp. 1394–1401.
10. [10] J. W. Taylor, P. E. McSharry, and R. Buizza, "Wind power density forecasting using ensemble predictions and time series models," IEEE Trans. Energy Convers., vol. 24, no. 3, pp. 775–782, 2009.
11. [11] M. Elsaraiti and A. Merabet, "A comparative analysis of the ARIMA and LSTM predictive models and their effectiveness for predicting wind speed," Energies, vol. 14, no. 20, p. 6782, 2021.
12. [12] N. L. M. Jailani et al., "Investigating the power of LSTM-based models in solar energy forecasting," Processes, vol. 11, no. 5, p. 1382, 2023.
13. [13] F. U. M. Ullah et al., "Short-term prediction of residential power energy consumption via CNN and multi-layer bidirectional LSTM networks," IEEE Access, vol. 8, pp. 123369–123380, 2019.
14. [14] S. Mahjoub, L. Chrifi-Alaoui, B. Marhic, and L. Delahoche, "Predicting energy consumption using LSTM, multi-layer GRU, and drop-GRU neural networks," Sensors, vol. 22, no. 11, p. 4062, 2022.
15. [15] M. Bilgili, N. Arslan, A. Şekertekin, and A. Yaşar, "Application of long short-term memory (LSTM) neural network based on deep learning for electricity energy consumption forecasting," Turk. J. Electr. Eng. Comput. Sci., vol. 30, no. 1, pp. 140–157, 2022.
16. [16] S. Arslan, "A hybrid forecasting model using LSTM and Prophet for energy consumption with decomposition of time series data," PeerJ Comput. Sci., vol. 8, p. e1001, 2022.
17. [17] A. Gasparin, S. Lukovic, and C. Alippi, "Deep learning for time series forecasting: The electric load case," CAAI Trans. Intell. Technol., vol. 7, no. 1, pp. 1–25, 2022.
18. [18] Akhil and Collaborator, Global Energy Statistics (1980-2021).
19. [19] G. P. Zhang, "Time series forecasting using a hybrid ARIMA and neural network model," Neurocomputing, vol. 50, pp. 159–175, 2003.
20. [20] L. Zhou, "Application of ARIMA model on prediction of China’s corn market," in J. Phys.: Conf. Ser., vol. 1941, no. 1, p. 012064, Jun. 2021.
21. [21] P. Manigandan et al., "Forecasting natural gas production and consumption in the United States—Evidence from SARIMA and SARIMAX models," Energies, vol. 14, no. 19, p. 6021, 2021.
22. [22] M. Baloch et al., "An intelligent SARIMAX-based machine learning framework for long-term solar irradiance forecasting at Muscat, Oman," Energies, vol. 17, no. 23, p. 6118, 2024.
23. [23] L. Zhou, Z. Luo, and X. Pan, "Machine learning-based system reliability analysis with Gaussian process regression," arXiv preprint arXiv:2403.11125, 2024.
24. [24] N. K. Manaswi and N. K. Manaswi, "RNN and LSTM," in Deep Learning with Applications Using Python: Chatbots and Face, Object, and Speech Recognition with TensorFlow and Keras, pp. 115–126, 2018.
25. [25] K. Albeladi, B. Zafar, and A. Mueen, "Time series forecasting using LSTM and ARIMA," Int. J. Adv. Comput. Sci. Appl., vol. 14, no. 1, pp. 313–320, 2023.
26. [26] J. Quinonero-Candela and C. E. Rasmussen, "A unifying view of sparse approximate Gaussian process regression," J. Mach. Learn. Res., vol. 6, pp. 1939–1959, 2005.
27. [27] M. Hossin and M. N. Sulaiman, "A review on evaluation metrics for data classification evaluations," Int. J. Data Min. Knowl. Manag. Process, vol. 5, no. 2, pp. 1–7, 2015.
28. [28] Accelerator Institute, Evaluating Machine Learning Models: Metrics and Techniques.
29. [29] F. R. Alharbi and D. Csala, “A Seasonal Autoregressive Integrated Moving Average with Exogenous Factors (SARIMAX) Forecasting Model-Based Time Series Approach,” Inventions 2022, Vol. 7, Page 94, vol. 7, no. 4, p. 94, Oct. 2022, doi: 10.3390/INVENTIONS7040094.
30. [30] M. Bilgili and E. Pinar, “Gross electricity consumption forecasting using LSTM and SARIMA approaches: A case study of Türkiye,” Energy, vol. 284, Dec. 2023, doi: 10.1016/J.ENERGY.2023.128575.
31. [31] B. Peak et al., “Peak Electrical Energy Consumption Prediction by ARIMA, LSTM, GRU, ARIMA-LSTM and ARIMA-GRU Approaches,” Energies 2023, Vol. 16, Page 4739, vol. 16, no. 12, p. 4739, Jun. 2023, doi: 10.3390/EN16124739.
32. [32] N. L. M. Jailani et al., “Investigating the Power of LSTM-Based Models in Solar Energy Forecasting,” Processes 2023, Vol. 11, Page 1382, vol. 11, no. 5, p. 1382, May 2023, doi: 10.3390/PR11051382.
33. [33] M. Sharifzadeh, A. Sikinioti-Lock, and N. Shah, “Machine-learning methods for integrated renewable power generation: A comparative study of artificial neural networks, support vector regression, and Gaussian Process Regression,” Renewable and Sustainable Energy Reviews, vol. 108, pp. 513–538, Jul. 2019, doi: 10.1016/J.RSER.2019.03.040.