Akıllı Şebeke Altyapısında Fotovoltaik Sistemlere Yönelik Siber Saldırıların Makine Öğrenmesi Yöntemleriyle Tespiti

Abstract

Günümüzde karbon emisyonlarının ve çevresel kaygıların artmasıyla birlikte, yenilenebilir enerji kaynaklarının güç sistemlerindeki payı da giderek artmaktadır. Meteorolojik koşullara bağlı olarak değişken güç üretimi gerçekleştiren bu sistemlerin enerji arz-talep dengesinde oluşturduğu dalgalanmalar, ancak akıllı şebeke altyapısıyla etkin bir şekilde yönetilebilmektedir. Akıllı şebekeler, haberleşme ve bilgi teknolojilerinin entegrasyonunu zorunlu kılarken, güç sistemlerini siber-fiziksel yapılara dönüştürerek yeni siber güvenlik risklerini de beraberinde getirmektedir. Dağıtık üretim kaynaklarının güç sistemine entegrasyonu, yeni siber güvenlik tehditlerini de beraberinde getirmektedir. Bu tehditlerin başında gelen sahte veri enjeksiyon saldırıları (False Data Injection Attacks- FDIA), durum tahminleyicilerini (State Estimators- SE) yanıltarak sistemde ciddi güvenlik açıklarına ve operasyonel risklere yol açabilmektedir. Bu çalışmada, fotovoltaik (PV) panellerden şebekeye aktarılan enerjinin manipüle edilmesi ve akıllı sayaç verilerinin yanıltılması yoluyla gerçekleştirilen siber saldırılar, makine öğrenmesi tabanlı ikili sınıflandırma yöntemleriyle analiz edilmiştir. Düşük, orta ve yüksek şiddetli siber saldırı senaryolarına göre değişen üretim miktarları, literatürde yaygın olarak kullanılan Rastegele Orman Algoritması (Random Forest Classifier- RFC), Aşırı Gradyan Artırma Algoritması (eXtreme Gradient Boosting Algorithm- XGBC) ve Gradyan Artırma Algoritması (Gradient Boosting Classifier- GBC) ile modellenmiştir ve yüksek doğruluk oranları elde edilmiştir. Modeller, düşük şiddetteki saldırı senaryosunda XGBC’den 92,33%, yüksek şiddetteki saldırı senaryosunda ise GBC’den 68,59% doğruluk oranı elde ederek yüksek doğruluk oranlarına ulaşmıştır.

Keywords

• Siber saldırı
• dağıtık üretim kaynakları
• makine öğrenmesi
• PV üretimi
• akıllı şebekeler

Detection of Cyberattacks on Photovoltaic Systems in Smart Grid Infrastructure Using Machine Learning Methods

Abstract

With the increasing concerns over carbon emissions and environmental sustainability, the share of renewable energy sources in power systems has been steadily rising. These systems, which generate variable power depending on meteorological conditions, cause fluctuations in the energy supply-demand balance. Such fluctuations can only be effectively managed through smart grid infrastructure. While smart grids necessitate the integration of communication and information technologies, they also transform power systems into cyber-physical structures, introducing new cybersecurity risks. The integration of distributed generation sources into power systems brings additional cybersecurity threats. Among these threats, false data injection attacks (FDIA) pose significant risks by misleading state estimators (SE), potentially creating severe security vulnerabilities and operational risks. In this study, cyberattacks aiming to manipulate the energy supplied to the grid from photovoltaic (PV) panels and to deceive smartmeter data were analyzed using machine learning-based binary classification methods. The variations in generation levels under low, medium, and high-intensity cyberattack scenarios were modeled using widely adopted algorithms in the literature, including Random Forest Classifier (RFC), XGBoost Classifier (XGBC), and Gradient Boosting Classifier (GBC). The models achieved high accuracy rates, with 92.33% obtained from XGBC in the low-severity attack scenario and 68.59% from GBC in the high-severity attack scenario.

Keywords

• Cyber-attack
• distributed energy resources
• machine learning
• PV generation
• smart grids

References

1. The International Renewable Energy Agency "IRENA", https://www.irena.org/Publications/2023/Jul/Renewableenergy-statistics-2023. Erişim tarihi: “05.03.2025”.
2. Fang X, Misra S, Xue G, Yang D. Smart Grid — The New and Improved Power Grid: A Survey. IEEE Commun Surv Tut 2012; 14(4): 944-980.
3. Guo L, Zhang J, Ye J, Coshatt SJ, Song W. Data-Driven Cyber-Attack Detection for PV Farms via Time-Frequency Domain Features. IEEE T Smart Grid 2022; 13(2): 1582-1597.
4. Ye J, et al. A Review of Cyber–Physical Security for Photovoltaic Systems. IEEE J Em Sel Top P 2022; 10(4): 4879-4901.
5. Nguyen T, Wang S, Alhazmi M, Nazemi M, Estebsari A, Dehghanian P. Electric Power Grid Resilience to Cyber Adversaries: State-of-the Art. IEEE Access 2020; 8: 87592-87608.
6. Eldahshan N, Asif M, Baajaj T, Shaaban MF, Osman AH, Tariq U. A new theft detection approach for cyberattacks in PV generation. 4th International Youth Conference on Radio Electronics, Electrical and Power Engineering (REEPE); 17-19 March 2022; Moscow, Russian Federation. 1-6.
7. Dehghanian P, Zhang B, Dokic T, Kezunovic M. Predictive Risk Analytics for Weather-Resilient Operation of Electric Power Systems. IEEE T Sustain Energ 2019; 10(1): 3-15.
8. Wei F, Wan Z, He H. Cyber-attack Recovery Strategy for Smart Grid Based on Deep Reinforcement Learning. IEEE T Smart Grid 2020; 11(3): 2476-2486.

9. Haimes YY. On the Definition of Resilience in Systems. Risk Analysis: An International Journal 2009; 29(4): 498-501.
10. Liu Y, Ning P, Reiter MK. False data injection attacks against state estimation in electric power grids. ACM T Inform Syst Secur 2011; 14(1): 1-33.
11. Chaojun G, Jirutitijaroen P, Motani M. Detecting False Data Injection Attacks in AC State Estimation. IEEE T Smart Grid 2015; 6(5): 2476-2483.
12. Ozay M, Esnaola I, Vural FTY, Kulkarni SR, Poor HV. Machine Learning Methods for Attack Detection in the Smart Grid. IEEE T Neur Net Lear Syst 2016; 27(8): 1773-1786.
13. Yu, J. J. Q., Hou, Y., & Li, V. O. K. Online False Data Injection Attack Detection with Wavelet Transform and Deep Neural Networks. IEEE T Ind Inform 2018; 14(7): 3271-3280.
14. Li F, Xie R, Wang Z, Guo L, Ye J, Ma P, Song WZ. Online Distributed IoT Security Monitoring with Multidimensional Streaming Big Data. IEEE Internet Things 2020; 7(5): 4387-4394.
15. Saiara SA, Ali MH. An ensemble learning based cyber attack detection technique for BESS integrated PV system. SoutheastCon; 15-24 March 2024; Atlanta, GA, USA. 392-397.
16. Riggs H, Tufail S, Khan M, Parvez I, Sarwat AI. Detection of false data injection of PV production. 2021 IEEE Green Technologies Conference (GreenTech); 7-9 April 2021; Denver, CO, USA. 7-12.
17. Ayad A, Farag HEZ, Youssef A, El-Saadany EF. Detection of false data injection attacks in smart grids using Recurrent Neural Networks. 2018 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT); 19-22 Feb. 2018; Washington, DC, USA. 1-5.
18. Zhao L, Li J, Li Q, Li F. A Federated Learning Framework for Detecting False Data Injection Attacks in Solar Farms. IEEE T Power Electr 2022; 37(3): 2496-2501.
19. Moradpour AM, Alizadeh MH, Delkhosh H. A new method based on symbolic regression to detect the probability of false data injection attacks on PV generation. 2023 13th Smart Grid Conference (SGC); 05-06 Dec. 2023; Tehran, Islamic Republic of Iran. 1-7.
20. Li Q, Zhang J, Ye J, Song W. Data-driven cyber-attack detection for photovoltaic systems: A transfer learning approach. 2022 IEEE Applied Power Electronics Conference and Exposition (APEC); 20-25 March 2022; Houston, TX, USA. 1926-1930.
21. Zhang J, Li Q, Ye J, Guo L. Cyber-physical security framework for Photovoltaic Farms. 2020 IEEE CyberPELS (CyberPELS); 13-13 Oct. 2020; Miami, FL, USA. 1-7.
22. Li Q, Li F, Zhang J, Ye J, Song W, Mantooth A. Data-driven cyberattack detection for photovoltaic (PV) systems through analyzing micro-PMU data. 2020 IEEE Energy Conversion Congress and Exposition (ECCE); 11-15 Oct. 2020; Detroit, MI, USA. 431-436.
23. Li F, Li Q, Zhang J, Kou J, Ye J, Song WZ, Mantooth HA. Detection and Diagnosis of Data Integrity Attacks in Solar Farms Based on Multilayer Long Short-Term Memory Network. IEEE T Power Electr 2021; 36(3): 2495-2498.
24. Zhang J, Guo L, Ye J, Giani A, Elasser A, Song W. Machine Learning-Based Cyber-Attack Detection in Photovoltaic Farms. IEEE Open J Power El 2023; 4: 658-673.
25. NASA Prediction Of Worldwide Energy Resources (POWER), http://www.ilo.org/global/topics/safety-and-healthatwork/lang--en/index.htm. Accessed: “14.01.2025”.
26. Masters GM, Renewable and Efficient Electric Power Systems, Wiley Interscience, 2013; 2nd ed. Hoboken, NJ, USA.
27. Maxeon Solar Technologies, https://sunpower.global/au/sites/default/files/2022-03/sp_max3_112c_blk_410-420_res_dc_ds_en_a4_544456.pdf. Accessed: “16.01.2025”.
28. Canadian Solar Power, https://www.canadiansolar.com/wp-content/uploads/2019/12/Canadian_Solar-Datasheet-HiKu_CS3L-P_EN.pdf. Accessed: “16.01.2025”.
29. JA Solar, https://www.jasolar.com/uploadfile/2021/0706/20210706053524693.pdf. Accessed: “16.01.2025”.
30. Parmar A, Katariya R, Patel V. A review on random forest: An ensemble classifier. International conference on intelligent data communication technologies and internet of things (ICICI); 07–08 Aug. 2018; Coimbatore, India. 758-763.
31. Scikit learn API, https://scikit-learn.org/stable/api/sklearn.ensemble.html/lang--en/index.htm. Accessed: “09.01.2025”.
32. Hossin M, Sulaiman M. A Review on Evaluation Metrics for Data Classification Evaluations. Int J Data Mın Model 2015; 5(2): 01-11.