Comparative Analysis of Machine and Deep Learning Methods in Estimating the Turkish Electricity Market Clearing Price

Öz

The estimation of the clearing price in the electricity market holds significant strategic importance within the energy sector. Energy firms can enhance their operational efficiency by providing clients with more dependable price alternatives through precise estimation of the market clearing price. The precise determination of the market clearing price holds significant significance in facilitating strategic decision-making for decision makers and investors operating within the energy sector. Accurate pricing projections are crucial for ensuring stability in the energy market and enhancing energy reliability for consumers. Hence, it is imperative to employ novel methodologies and enhance the precision of predictions within the energy sector in order to ascertain precise price estimates. This study utilized hourly power data derived from various sources such as natural gas, dam, lignite, imported coal, wind, solar, geothermal, and biomass. Additionally, hourly electricity demand data was employed as input variables to estimate the clearing price of the electricity market. The study encompasses a total of 8772 hours of data collected between April 17, 2023, to April 16, 2023. The study employed linear regression, XGBoost, Random Forest, LSTM, and SVR techniques for prediction. The models were evaluated by comparing their performances using statistical coefficients such as RMSE, MSE, MAE, and R2. Based on the acquired performance measures, it was noted that the XGBoost approach exhibited the highest level of prediction performance.

Anahtar Kelimeler

• Electricity clearing price
• machine learning
• deep learning
• decision support
• energy.

Türkiye Elektrik Piyasası Takas Fiyatının Tahmininde Makine ve Derin Öğrenme Yöntemlerinin Karşılaştırmalı Analizi

Öz

Elektrik piyasa takas fiyatının tahmini enerji alanında stratejik öneme sahiptir. Doğru bir şekilde piyasa takas fiyatının tahmin edilmesi ile enerji şirketleri müşterilerine daha güvenilir fiyat alternatifleri sunarak operasyonel verimliliğini artırabilmektedir. Piyasa takas fiyatının doğru bir şekilde tahmini enerji sektöründeki karar vericilerin ve yatırımcıların stratejik seçimler yapmalarına yardımcı olması açısından büyük önem taşımaktadır. Enerji piyasasında istikrarın sağlanması ve tüketiciler açısından enerji güvenilirliğini artırmak için fiyat tahminlerinin doğru bir şekilde yapılması gerekmektedir. Bu nedenle enerji endüstrisinde doğru fiyat tahminlerinin tapılması için yeni yöntemlerin kullanılması ve daha doğru tahminlerin yapılması oldukça önemlidir. Bu çalışmada elektrik piyasa takas fiyatının tahmin edilmesi için Doğalgaz, baraj, linyit, ithal kömür, rüzgâr, güneş, jeotermal ve biokütleden üretilen saatlik elektrik verileri ile saatlik elektrik talep verileri girdi değişkeni olarak kullanılmıştır. Çalışma 17.04.2023-16.04.2024 arasındaki 8772 saatlik veriyi kapsamaktadır. Çalışmada XGBoost, Random Forest, LSTM ve SVR yöntemlerinin yanı sıra doğrusal regresyon ile de tahmin yapılmıştır. Modellerin performansları RMSE, MSE, MAE ve R2 istatistik katsayıları kullanılarak karşılaştırılmıştır. Elde edilen performans metriklerine göre en iyi tahmin performansının XGBoost yöntemi tarafından üretildiği gözlemlenmiştir.

Anahtar Kelimeler

• Piyasa takas fiyatı
• makine öğrenmesi
• derin öğrenme
• karar destek
• enerji.

Kaynakça

1. Haliloğlu EY, Tutu BE. Türkiye için kısa vadeli elektrik enerjisi talep tahmini. Yasar University EJ; 2018; 13(51): 243-255.
2. Nebati EE, TAŞ M, Ertaş G. Türkiye’de elektrik tüketiminde talep tahmini: zaman serisi ve regresyon analizi ile karşılaştırma. Eur J Sci Technol; 2021; (31): 348-357.
3. Contreras J, Espínola R, Nogales F, Conejo A. Arima models to predict next-day electricity prices. IEEE Trans Power Syst 2003; 18(3): 1014-1020.
4. Amjady N, Daraeepour A, Keynia F. Day-ahead electricity price forecasting by modified relief algorithm and hybrid neural network. IET Gener Transm Distrib; 2010; 4(3): 432-444.
5. Carpio KJE, Go AML, Roncal CKM. Forecasting day-ahead electricity prices of Singapore through ARIMA and wavelet-ARIMA. DLSU Bus Econ Rev; 2012; 22(1): 97-118.
6. Voronin S, Partanen J, Kauranne T. A hybrid electricity price forecasting model for the Nordic electricity spot market. Int Trans Electr Energy Syst; 2013; 24(5): 736-760.
7. Wang Z, Liu F, Wu J, Wang J. A hybrid forecasting model based on bivariate division and a backpropagation artificial neural network optimized by chaos particle swarm optimization for day-ahead electricity price. Abstr Appl Anal 2014; 2014: 1-31.
8. Jiang P, Liu F, Song Y. A hybrid multi-step model for forecasting day-ahead electricity price based on optimization, fuzzy logic and model selection. Energies 2016; 9(8): 618.

9. Gao G, Lo K, Lu J. Risk assessment due to electricity price forecast uncertainty in UK electricity market. In: 52nd International Universities Power Engineering Conference (UPEC); 2017; New York, NY, USA: IEEE. pp. 1-6.
10. Pourdaryaei A, Mokhlis H, Illias H, Kaboli S, Ahmad S. Short-term electricity price forecasting via hybrid backtracking search algorithm and ANFIS approach. IEEE Access 2019; 7: 77674-77691.
11. Huang C, Shen Y, Chen Y, Chen H. A novel hybrid deep neural network model for short-term electricity price forecasting. Int J Energy Res; 2020; 45(2): 2511-2532.
12. Karatekin C, Başaran T. Forecasting the day ahead electricity energy price by using data analysis methods. Iğdır Univ J Inst Sci Technol; 2022; 12(4): 2075-2084.
13. Arslan B, Ertuğrul İ. Çoklu regresyon, ARIMA ve yapay sinir ağı yöntemleri ile Türkiye elektrik piyasasında fiyat tahmin ve analizi. J Manag Econ Res; 2022; 20(1): 331-353.
14. Misiorek A, Trueck S, Weron R. Point and interval forecasting of spot electricity prices: linear vs. non-linear time series models. Stud Nonlinear Dyn Econom; 2006; 10(3).
15. Wang R, Fu-xiong W, Ji W. Particle swarm optimization based GM(1,2) method on day-ahead electricity price forecasting with predicted error improvement. In: 2nd International Workshop on Database Technology and Applications; 2010; Wuhan, China. pp. 1-6.
16. Martínez-Álvarez F, Troncoso A, Riquelme J, Aguilar-Ruiz J. Energy time series forecasting based on pattern sequence similarity. IEEE Trans Knowl Data Eng 2011; 23(8): 1230-1243.
17. Ozozen A, Kayakutlu G, Ketterer M, Kayalica O. A combined seasonal ARIMA and ANN model for improved results in electricity spot price forecasting: case study in Turkey. In: Proceedings of Portland International Conference on Management of Engineering and Technology (PICMET); 2016; Portland, OR, USA. pp. 2681-2690.
18. Costa e Silva E., Borges A., Teodoro MF., Andrade MA., Covas R. Time series data mining for energy prices forecasting: an application to real data. In: 16th International Conference on Intelligent Systems Design and Applications (ISDA 2016); 2016; Porto, Portugal: Springer International Publishing. pp. 649-658.
19. Li W, Becker DM. Day-ahead electricity price prediction applying hybrid models of LSTM-based deep learning methods and feature selection algorithms under consideration of market coupling. Energy 2021; 237: 121543.
20. Tschora L, Pierre E, Plantevit M, Robardet C. Electricity price forecasting on the day-ahead market using machine learning. Appl Energy 2022; 313: 118752.
21. Memarzadeh G, Keynia F. Short-term electricity load and price forecasting by a new optimal LSTM-NN based prediction algorithm. Electr Power Syst Res; 2021; 192: 106995.
22. Zhang J, Tan Z, Wei Y. An adaptive hybrid model for short term electricity price forecasting. Appl Energy 2020; 258: 114087.
23. Zhang R, Li G, Ma Z. A deep learning based hybrid framework for day-ahead electricity price forecasting. IEEE Access 2020; 8; 143423-143436.
24. Qiao W, Yang Z. Forecast the electricity price of US using a wavelet transform-based hybrid model. Energy; 2020; 193: 116704.
25. Huang CJ, Shen Y, Chen YH, Chen HC. A novel hybrid deep neural network model for short-term electricity price forecasting. Int J Energy Res; 2021; 45(2): 2511-2532.
26. Var H, Türkay BE. Yapay sinir ağları kullanılarak kısa dönem elektrik yükü tahmini short term electric load forecasting using artificial neural networks. In: Elektrik–Elektronik–Bilgisayar ve Biyomedikal Mühendisliği Sempozyumu; 2014; Bursa, Türkiye. pp. 34-37.
27. Kalfa VR, Arslan B, Ertuğrul İ. Determining the factors affecting the market clearing price by using multiple linear regression method. Alphanumeric J; 2021; 9(1): 35-48.
28. Demirezen S, Çetin M. Rassal Orman Regresyonu ve Destek Vektör Regresyonu ile Piyasa Takas Fiyatının Tahmini. J. Quant Sci; 2021; 3(1): 1-15.
29. Chen T, Guestrin C. XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; 2016; New York, NY, USA. pp. 785-794.
30. Jabeur SB, Mefteh-Wali S, Viviani JL. Forecasting gold price with the XGBoost algorithm and SHAP interaction values. Ann Oper Res; 2024; 334(1): 679-699.
31. Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput 1997; 9(8): 1735-1780.
32. Zhou F, Huang Z, Zhang C. Carbon price forecasting based on CEEMDAN and LSTM. Appl Energy 2022; 311: 118601.
33. Chen W, Li Y, Xue W, Shahabi H, Li S, Hong H, Wang X, Bian H, Zuang S, Pradhan BB, Ahmad BB. Modeling flood susceptibility using data-driven approaches of naïve bayes tree, alternating decision tree, and random forest methods. Sci Total Environ; 2020; 701: 134979.
34. Schonlau M, Zou RY. The random forest algorithm for statistical learning. Stata J; 2020; 20(1): 3-29.
35. Cortes C, Vapnik V. Support vector networks. Mach Learn; 1995; 20: 273-297.
36. Vapnik VN. An overview of statistical learning theory. IEEE Trans. Neural Netw.; 1999; 10(5): 988-999.
37. Smola AJ, Schölkopf B. A tutorial on support vector regression. Stat Comput; 2004; 14: 199-222.
38. Zouzou Y, Citakoglu H. Reference evapotranspiration prediction from limited climatic variables using support vector machines and Gaussian processes. Eur J Sci Technol; 2021; 28: 346-351.
39. Arifoğlu A, Kandemir T. Electricity price forecasting in Turkish day-ahead market via deep learning techniques. Mehmet Akif Ersoy Univ J Econ Admin Sci; 2022; 9(2): 1433.