Açıklanabilir Makine Öğrenmesi Tabanlı Elektrikli Araç Entegrasyon Kapasitesi Değerlendirmesi

Öz

Elektrikli araçların yaygınlaşması, dağıtım sistemlerinde gerilim profili bozulmaları ve aktif güç kayıplarında artış gibi işletme sorunlarının ayrıntılı biçimde değerlendirilmesini gerektirmektedir. Bu çalışmada, elektrikli araç entegrasyonunun dağıtım sistemi performansı üzerindeki etkileri açıklanabilir makine öğrenmesi yaklaşımıyla incelenmiştir. IEEE 33 bara dağıtım sistemi MATPOWER ortamında modellenmiş ve Norveç’teki konut kullanıcılarına ait gerçek saatlik şarj verileri kullanılarak 20.000 işletme senaryosu oluşturulmuştur. Farklı penetrasyon seviyeleri, bağlantı noktaları, yük seviyeleri ve güç faktörleri dikkate alınmıştır. Minimum bara gerilimi ile aktif güç kayıplarının tahmininde CatBoost, XGBoost, LightGBM ve Random Forest algoritmaları karşılaştırılmıştır. CatBoost her iki çıktı için en yüksek doğruluğu sağlamıştır. Açıklanabilirlik analizi, sistem yükü, toplam şarj gücü ve penetrasyon seviyesinin en etkili değişkenler olduğunu göstermiştir. Bağlantı noktalarına göre hesaplanan entegrasyon kapasiteleri, konum seçiminin şebeke performansında belirleyici olduğunu ortaya koymuştur. Önerilen yaklaşım, planlama çalışmaları için hızlı, veri odaklı ve yorumlanabilir bir karar destek çerçevesi sunmaktadır.

Anahtar Kelimeler

• Elektrikli Araçlar
• Dağıtım Sistemleri
• Entegrasyon Kapasitesi
• Makine Öğrenmesi
• SHAP Analizi

Destekleyen Kurum

Bu çalışma herhangi bir kurum veya kuruluş tarafından finansal olarak desteklenmemiştir.

Etik Beyan

Bu çalışma yalnızca açık erişimli, anonimleştirilmiş veriler ve bilgisayar tabanlı benzetim sonuçları kullanılarak gerçekleştirilmiştir. Çalışma kapsamında insan veya hayvan denek kullanılmadığından etik kurul izni gerekmemektedir. Yazarlar, çalışmanın araştırma ve yayın etiğine uygun olarak hazırlandığını beyan ederler.

Teşekkür

Bulunmamaktadır.

Explainable Machine Learning-Based Assessment of Electric Vehicle Hosting Capacity in Distribution Systems

Öz

The adoption of electric vehicles requires assessment of operational problems such as voltage deterioration and increased active power losses in distribution systems. This study investigates electric vehicle integration effects through an explainable machine learning approach. The IEEE 33-bus distribution system was modeled in MATPOWER, and 20,000 operating scenarios were generated using real hourly charging data from residential users in Norway. Different penetration levels, connection points, load levels, and power factors were considered. CatBoost, XGBoost, LightGBM, and Random Forest algorithms were compared for predicting minimum bus voltage and active power losses. CatBoost achieved the highest accuracy for both outputs. The explainability analysis showed that system loading, total charging power, and penetration level were the most influential variables. Hosting capacities calculated for different connection points demonstrated that location selection has a decisive effect on network performance. The proposed approach provides a fast, data-driven, and interpretable decision-support framework for electric vehicle integration planning.

Anahtar Kelimeler

• Electric Vehicles
• Distribution Systems
• Hosting Capacity
• Machine Learning
• SHAP Analysis

Destekleyen Kurum

This study received no financial support from any institution or organization.

Etik Beyan

This study was conducted solely using publicly available anonymized data and computer-based simulation results. Since no human participants or animals were involved in the study, ethics committee approval was not required. The authors declare that the study was prepared in accordance with research and publication ethics.

Teşekkür

None

Kaynakça

1. S. Rahman, I. A. Khan, A. A. Khan, A. Mallik, and M. F. Nadeem, “Comprehensive review and impact analysis of integrating projected electric vehicle charging load to the existing low-voltage distribution system,” Renew. Sustain. Energy Rev., vol. 153, Art. no. 111756, 2022.
2. S. Ray, K. Kasturi, S. Patnaik, and M. R. Nayak, “Review of electric vehicles integration impacts in distribution networks: Placement, charging/discharging strategies, objectives and optimisation models,” J. Energy Storage, vol. 72, Art. no. 108672, 2023.
3. N. Panossian, M. Muratori, B. Palmintier, A. Meintz, T. Lipman, and K. Moffat, “Challenges and opportunities of integrating electric vehicles in electricity distribution systems,” Curr. Sustain./Renew. Energy Rep., vol. 9, no. 2, pp. 27–40, 2022.
4. M. S. H. Nizami, M. J. Hossain, and K. Mahmud, “Coordinated electric vehicle management system for grid support services in residential networks,” IEEE Syst. J., vol. 15, no. 2, pp. 2066–2077, 2021.
5. A. Tayri and X. Ma, “Grid impacts of electric vehicle charging: A review of challenges and mitigation strategies,” Energies, vol. 18, no. 14, Art. no. 3807, 2025.
6. T. Yuvaraj, K. R. Devabalaji, J. A. Kumar, S. B. Thanikanti, and N. I. Nwulu, “A comprehensive review and analysis of the allocation of electric vehicle charging stations in distribution networks,” IEEE Access, vol. 12, pp. 5404–5461, 2024.
7. C. de Mattos Affonso and M. Kezunovic, “Technical and economic impact of PV-BESS charging station on transformer lifetime: A case study,” IEEE Trans. Smart Grid, vol. 10, no. 4, pp. 4683–4692, 2019.
8. D. Alame, M. Azzouz, and N. Kar, “Assessment and mitigation of electric vehicle harmonic currents on distribution systems,” Energies, vol. 13, no. 12, Art. no. 3257, 2020.

9. Z. M. Zenhom, S. H. A. Aleem, A. F. Zobaa, and T. A. Boghdady, “A comprehensive review of renewables and electric vehicles hosting capacity in active distribution networks,” IEEE Access, vol. 12, pp. 3672–3699, 2024.
10. S. Fatima, V. Püvi, M. Lehtonen, and M. Pourakbari-Kasmaei, “A review of electric vehicle hosting capacity quantification and improvement techniques for distribution networks,” IET Gener. Transm. Distrib., vol. 18, no. 6, pp. 1095–1113, 2024.
11. M. D. Kamruzzaman, X. Zhang, M. Abdelmalak, M. Benidris, and D. Shi, “A method to evaluate the maximum hosting capacity of power systems to electric vehicles,” in Proc. Int. Conf. Probabilistic Methods Applied to Power Systems (PMAPS), Liège, Belgium, 2020.
12. S. Wang, C. Li, Z. Pan, and J. Wang, “Probabilistic method for distribution network electric vehicle hosting capacity assessment based on combined cumulants and Gram–Charlier expansion,” Energy Procedia, vol. 158, pp. 5067–5072, 2019.
13. M. Al-Saffar, S. Zhang, A. Nassif, and P. Musilek, “Assessment of photovoltaic hosting capacity of existing distribution circuits,” in Proc. IEEE Canadian Conf. Electrical and Computer Engineering (CCECE), Edmonton, Canada, pp. 1–4, 2019.
14. M. Kamruzzaman and M. Benidris, “Reliability-based metrics to quantify the maximum permissible load demand of electric vehicles,” IEEE Trans. Ind. Appl., vol. 55, no. 4, pp. 3365–3375, 2019.
15. A. E. Avila-Rojas, P. M. De Oliveira-De Jesus, and M. Alvarez, “Distribution network electric vehicle hosting capacity enhancement using an optimal power flow formulation,” Electr. Eng., vol. 104, no. 3, pp. 1337–1348, 2022.
16. V. Umoh, A. Adebiyi, and K. Moloi, “Hosting capacity assessment of South African residential low-voltage networks for electric vehicle charging,” Eng, vol. 4, no. 3, pp. 1965–1980, 2023.
17. O. Karahan and M. Bağrıyanık, “Machine learning-based dynamic hosting capacity for bidirectional EVs,” IEEE Access, vol. 14, pp. 30906–30915, 2026.
18. S. Deshmukh, A. Iqbal, K. Suburna, K. Thomas, A. Rahman, and S. Islam, “Evaluation of electric vehicles hosting capacity using artificial intelligence techniques,” in Proc. 19th IEEE Int. Conf. Compatibility, Power Electronics and Power Engineering, Antalya, Türkiye, pp. 1–6, 2025.
19. N. Hariyanto, A. A. R. Leksono, and K. M. Banjarnahor, “A machine learning framework for stochastic assessment of combined PV/EV hosting capacity boundaries in MV distribution networks,” SSRN, pp. 1–14, 2025.
20. M. Asadi, K. Jalilpoor, R. Claeys, and J. Desmet, “Machine learning-based model to estimate the dynamic hosting capacity in distribution network,” in Proc. IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe), pp. 1–5, 2024.
21. M. E. Baran and F. F. Wu, “Network reconfiguration in distribution systems for loss reduction and load balancing,” IEEE Trans. Power Deliv., vol. 4, no. 2, pp. 1401–1407, 1989.
22. R. D. Zimmerman, C. E. Murillo-Sánchez, and R. J. Thomas, “MATPOWER: Steady-state operations, planning, and analysis tools for power systems research and education,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 12–19, 2011.
23. A. L. Sorensen, “Electric vehicle charging dataset with 35,000 charging sessions from 12 residential locations in Norway,” Zenodo, 2024.
24. R. Zhu, G. Ciren, B. Tang, et al., “Power system short-term voltage stability assessment based on improved CatBoost with consideration of model confidence,” Energy Sci. Eng., vol. 11, no. 2, pp. 783–795, 2023.
25. M. Chen, Q. Liu, S. Chen, et al., “XGBoost-based algorithm interpretation and application on post-fault transient stability status prediction of power system,” IEEE Access, vol. 7, pp. 13149–13158, 2019.
26. Y. Yang, Y. Wang, and X. Zhang, “Research on power system small-signal stability analysis and correction based on LightGBM algorithm,” Electr. Eng., vol. 106, no. 4, pp. 4469–4486, 2024.
27. Z. Mustaffa and M. H. Sulaiman, “Random forest-based wind power prediction method for sustainable energy system,” Cleaner Energy Syst., Art. no. 100210, 2025.
28. D. Chicco, M. J. Warrens, and G. Jurman, “The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE, and RMSE in regression analysis evaluation,” PeerJ Comput. Sci., vol. 7, Art. no. e623, 2021.
29. S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” Advances in Neural Information Processing Systems, vol. 30, pp. 4765–4774, 2017.