Elektrikli Araç Büyümesine Dayalı Elektrik Talebi Tahmini: Gradient Boosting, KNN ve Extra Trees Modellerinin Karşılaştırmalı Analizi

Öz

Bu araştırmada, elektrikli araç sayısının yıllık elektrik tüketimi üzerindeki etkisi makine öğrenmesi yöntemleri kullanılarak incelenmiştir. Güncel veriler kullanılarak, elektrik tüketimi tahmini için Gradient Boosting, K-Nearest Neighbors (KNN), Extra Trees, Bayesian Ridge ve Elastic Net modelleri uygulanmıştır. Modeller, MAE (Ortalama Mutlak Hata), RMSE (Kök Ortalama Kare Hata) ve R² (Determinasyon Katsayısı) gibi metriklerle değerlendirilmiştir. Elde edilen sonuçlara göre, en başarılı model KNN Regressor olmuştur. KNN modeli MAE = 17,186, RMSE = 21,003 ve R² = 0.81 değeriyle en yüksek doğruluğa ulaşmıştır. Gradient Boosting ve Extra Trees modelleri de sırasıyla 24,283 ve 22,965 RMSE değerleriyle rekabetçi sonuçlar üretmiştir. Buna karşılık, Bayesian Ridge modeli -0.04 R² skoru ile yetersiz bir performans göstermiştir, yani model veri setindeki ilişkileri başarılı bir şekilde öğrenememiştir. Elastic Net modeli ise RMSE = 33,064 ve R² = 0.52 ile orta seviyede bir başarı sergilemiştir. Bu sonuçlar, elektrikli araç sayısı ile elektrik tüketimi arasında güçlü ancak doğrusal olmayan bir ilişki olduğunu göstermektedir. Özellikle KNN gibi parametrik olmayan modellerin en iyi performansı göstermesi, elektrikli araçların enerji tüketimine olan etkisinin karmaşık ve doğrudan doğrusal olmayan bir yapı sergilediğini kanıtlamaktadır. Bu sonuç, gelecekte elektrikli araç sayısının artmasıyla elektrik talebinde de orantılı bir yükseliş olacağını ve enerji altyapısının bu doğrultuda planlanması gerektiğini göstermektedir. Özellikle şehirlerarası şarj istasyonlarının artırılması, yenilenebilir enerji kaynaklarının entegrasyonu ve akıllı şebeke yönetimi gibi stratejiler, bu büyüyen talebe uyum sağlamak için kritik öneme sahip olacaktır.

Anahtar Kelimeler

• Elektrikli Araçlar
• Makine Öğrenmesi
• Şehir Enerji Talep Planlaması

Electricity Demand Forecasting Based on Electric Vehicle Growth: A Comparative Analysis of Gradient Boosting, KNN, and Extra Trees Models

Öz

In this study, the impact of the number of electric vehicles on annual electricity consumption was examined using machine learning methods. Using up-to-date data, Gradient Boosting, K-Nearest Neighbors (KNN), Extra Trees, Bayesian Ridge, and Elastic Net models were applied for electricity consumption forecasting. The models were evaluated using metrics such as MAE (Mean Absolute Error), RMSE (Root Mean Square Error), and R² (Coefficient of Determination). According to the results obtained, the most successful model was the KNN Regressor. The KNN model achieved the highest accuracy with MAE = 17.186, RMSE = 21.003, and R² = 0.81. Gradient Boosting and Extra Trees models also produced competitive results, with RMSE values of 24.283 and 22.965, respectively. In contrast, the Bayesian Ridge model exhibited poor performance with an R² score of -0.04, indicating that it failed to successfully learn the relationships in the dataset. The Elastic Net model demonstrated moderate success with RMSE = 33.064 and R² = 0.52. These results indicate a strong but nonlinear relationship between the number of electric vehicles and electricity consumption. The fact that non-parametric models such as KNN performed best proves that the impact of electric vehicles on energy consumption follows a complex and inherently nonlinear pattern. This finding suggests that as the number of electric vehicles increases in the future, electricity demand will rise proportionally, necessitating appropriate planning of the energy infrastructure. Strategies such as expanding intercity charging stations, integrating renewable energy sources, and implementing smart grid management will be critical in adapting to this growing demand.

Anahtar Kelimeler

• Electric Vehicles
• Machine Learning
• Urban Energy Demand Planning

Kaynakça

1. ACEA. Interactive map – Electric vehicle purchase incentives per country in Europe; 2020.
2. Al-Dhaifallah M, Refaat MM, Alaas Z, Abdel Aleem SHE, El-Kholy EE, Ali ZM. (2024) Multi- objectives transmission expansion planning considering energy storage systems and high penetration of renewables and electric vehicles under uncertain conditions. Energy Rep;11:4143–64.
3. Amalou Ibtissam, Mouhni Naoual, Abdali Abdelmounaim. (2022) Multivariate time series prediction by RNN architectures for energy consumption forecasting. Energy Rep;8:1084–91. Bain & Company; (2019) Predicting the Tipping Point for Electric Vehicles. .
4. Cover, T., & Hart, P. (1967). Nearest neighbor pattern classification. IEEE transactions on information theory, 13(1), 21-27. Eagon Matthew J, Kindem Daniel K, Panneer Selvam Harish, Northrop William F. (2022) Neural network-based electric vehicle range prediction for smart charging optimization. J Dyn Syst Meas Control;144(1):011110.
5. EU Commission, 2020. 2030 Climate & Energy Framework. . Friedman, J. H. (2001), Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189-1232. Fu Z, Dong P, Ju Y. (2020) An intelligent electric vehicle charging system for new energy companies based on consortium blockchain. J. Cleaner Prod.;261:121219.
6. Geurts, P., Ernst, D., & Wehenkel, L. (2006). Extremely randomized trees. Machine learning, 63, 3- 42. Greene DL, Park S, Liu C. (2014) Analyzing the transition to electric drive vehicles in the U.S. Futures;58:34–52.
7. Hudson B, Razeghi G, Samuelsen S. (2024) Mitigating impacts associated with a high- penetration of plug-in electric vehicles on local residential smart grid infrastructure. J Power Sources;593:233961 Huang Xingshuai, Wu Di, Boulet Benoit. (2020) Ensemble learning for charging load forecasting of electric vehicle charging stations. In: 2020 IEEE electric power and energy conference. EPEC, IEEE;, p. 1–5.
8. IEA - Tracking Transport; 2020. n.d. Jlifi Boutheina, Medini Mahdi, Duvallet Claude. (2024) A guided genetic algorithm-based ensemble voting of polynomial regression and LSTM (GGA-PolReg-LSTM) for congestion prediction using IoT and air quality data in sustainable cities. J Supercomput;1–41.

9. K. N. K., Jayalakshmi NS, Jadoun VK. (2024) Risk-based dynamic pricing by metaheuristic optimization approach for electric vehicle charging infrastructure powered by grid integrated microgrid system. Electr. Power Syst. Res.;230:110250.
10. Karaoğlu, A., & Söyler, H. (2025). A New Approach for Improving Biodiesel Conversion Efficiency: A Stacking Ensemble Model Based on Linear Regression Approach with GAN-Enhanced. Arabian Journal for Science and Engineering, 1-21. Lestaluhu Said, Baharuddin Tawakkal, Wance Marno (2023) Indonesian policy campaign for electric vehicles to tackle climate change: Maximizing social media. Int J Sustain Dev Plan;18(8).
11. L´ evay PZ, Drossinos Y, Thiel C. (2017) The effect of fiscal incentives on market penetration of electric vehicles: a pairwise comparison of total cost of ownership. Energy Policy;105:524–33.
12. Mamoudan Mobina Mousapour, Jafari Ali, Mohammadnazari Zahra, Nasiri Mohammad Mahdi, Yazdani Maziar. (2023) Hybrid machine learning-metaheuristic model for sustainable agri-food production and supply chain planning under water scarcity. Resour Environ Sustain;14:100133.
13. Malik Hasmat, Alotaibi Majed A, Almutairi Abdulaziz. (2022) A new hybrid model combining EMD and neural network for multi-step ahead load forecasting. J Intell Fuzzy Systems;42(2):1099–114. Mersky AC, Sprei F, Samaras C, Qian Z(Sean) (2016) Effectiveness of incentives on electric vehicle adoption in Norway. Transp Res Part D Transp Environ;46: 56–68.
14. Pham Anh-Duc, Ngo Ngoc-Tri, Truong Thi Thu Ha, Huynh Nhat-To, Truong Ngoc-Son. (2020) Predicting energy consumption in multiple buildings using machine learning for improving energy efficiency and sustainability. J Clean Prod;260:121082.
15. Pierrou G, Valero-De la Flor C, Hug G. (2024) Optimal EV charging scheduling at electric railway stations under peak load constraints. Elect. Power Syst. Res.;235: 110612
16. Qian K, Fachrizal R, Munkhammar J, Ebel T, Adam R. (2024) Large-scale EV charging scheduling considering on-site PV generation by combining an aggregated model and sorting-based methods. Sustainable Cities Soc;107:105453.
17. Qian K, Zhou C, Yuan Y. (2015) Impacts of high penetration level of fully electric vehicles charging loads on the thermal ageing of power transformers. Int. J. Elect. Power Energy Syst.;65:102–12. Requia WJ, Mohamed M, Higgins CD, Arain A, Ferguson M. (2018) How clean are electric vehicles? Evidence-based review of the effects of electric mobility on air pollutants, greenhouse gas emissions and human health. Atmos Environ;185:64–77.
18. Rüdisüli M, Bach C, Bauer C, Beloin-Saint-Pierre D, Elber U, Georges G, et al. (2022) Prospective life-cycle assessment of greenhouse gas emissions of electricity-based mobility options. Appl Energy;306:118065. Shanmuganathan Jaikumar, Victoire Aruldoss Albert, Balraj Gobu, Victoire Amalraj. (2022) Deep learning LSTM recurrent neural network model for prediction of electric vehicle charging demand. Sustainability;14(16):10207.
19. Sierzchula W, Bakker S, Maat K, van Wee B. (2014) The influence of financial incentives and other socio-economic factors on electric vehicle adoption. Energy Policy; 68:183–94.
20. Stoean Catalin, Zivkovic Miodrag, Bozovic Aleksandra, Bacanin Nebojsa, Strulak-Wójcikiewicz Roma, Antonijevic Milos, Stoean Ruxandra. (2023) Metaheuristic-based hyperparameter tuning for recurrent deep learning: Application to the prediction of solar energy generation. Axioms;12(3):266.
21. Sun D., Zheng Y, Duan R. (2021) Energy consumption simulation and economic benefit analysis for urban electric commercial-vehicles. Transp Res Part D Transp Environ;101:103083.
22. Tipping, M. E. (2001). Sparse Bayesian learning and the relevance vector machine. Journal of machine learning research, 1(Jun), 211-244. Transport emissions; 2017. .
23. UN, United Nations. Adoption of the Paris Agreement Online; 2015. (22.12.2015). n.d. Wang Y, Ning W, Zhang S, Yu H, Cen H, Wang S. (2021) Architecture and key terminal technologies of 5G-based internet of vehicles. Comput Electr Eng;95:107430.
24. Wang Jian Qi, Du Yu, Wang Jing. (2020) LSTM based long-term energy consumption prediction with periodicity. Energy;197:117197. Yang Yu, tianlei Z, zhichao L. (2021) Comparative analysis of energy consumption standards based on electric passenger vehicle economic simulation platform. J Phys: Conf Ser;2022(1):012027.
25. Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society Series B: Statistical Methodology, 67(2), 301-320.