Real-Time Fault Risk Forecasting on Transmission Lines Using Hybrid Deep-Learning Models and Generative Data Augmentation

The decarbonization of electric power systems and the increasing penetration of inverterbased resources have altered transmission system dynamics, reducing effective inertia and narrowing operational margins. Under these conditions, transmission line faults and cascading outages may develop faster and with weaker precursors than in traditionally synchronous systems. Conventional protection and monitoring schemes, largely based on fixed thresholds of megawatt (MW) flow or fault current magnitude, are reactive and provide limited lead time for preventive or corrective operator actions. The proposed forecasting approach improves situational awareness by detecting early deviations in PMU-derived system dynamics.