Operation-adversarial scenario generation

This paper proposes a modified conditional generative adversarial network (cGAN) model to generate net load scenarios for power systems that are statistically credible, conditioned by given labels (e.g., seasons), and, at the same time, “stressful” system operations and dispatch decisions. The measure of stress used in this is based on operating cost increases due changes. proposed operation-adversarial cGAN (OA-cGAN) internalizes DC optimal flow seeks maximize achieve worst-case data generation. training testing stages employed OA-cGAN use historical day-ahead forecast errors has been implemented realistic NYISO 11-zone system. Our numerical experiments demonstrate generated lead more cost-effective reliable