Physics-aware learning for detecting robust universal perturbation attacks in wind power forecasting

Key Points

• The primary outcome shows enhanced robustness in detecting universal perturbation attacks, improving forecasting accuracy.
• Increased detection rates of approximately 85% demonstrate the effectiveness of physics-aware learning in mitigating risks.
• Observational analysis of machine learning models illustrates significant improvements in forecasting under adversarial conditions.
• Improving robustness against such attacks is crucial for maintaining reliable energy forecasts in wind power systems.