Analysis of deceptive data attacks with adversarial machine learning for solar photovoltaic power generation forecasting

Kuzlu M., Sarp S., Catak F. O., Cali U., Zhao Y., ELMA O., ...Daha Fazla

Electrical Engineering, cilt.106, sa.2, ss.1815-1823, 2024 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 106 Sayı: 2
  • Basım Tarihi: 2024
  • Doi Numarası: 10.1007/s00202-022-01601-9
  • Dergi Adı: Electrical Engineering
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Compendex, INSPEC, DIALNET
  • Sayfa Sayıları: ss.1815-1823
  • Anahtar Kelimeler: Adversarial machine learning attacks, Forecasting, Solar PV energy generation forecasting
  • Yozgat Bozok Üniversitesi Adresli: Hayır

Özet

The solar photovoltaics (PV) energy resources have become more important with their significant contribution to the current power grid among renewable energy resources. However, the integration of the solar PV causes reliability issues in the power grid due to its high dependence on the weather condition. The predictability and stability of forecasting are critical for fully utilizing solar power. This study presents an Artificial Neural Network (ANN)-based solar PV power generation forecasting using a public dataset to form a basis experimental testbed to demonstrate analysis and impact of deceptive data attacks with adversarial machine learning. In addition, it evaluates the algorithms’ performance using the Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and Mean Average Error (MAE) metrics for two main cases, i.e., with and without adversarial machine learning attacks. The results show that the ANN-based models are vulnerable to adversarial attacks.