Deep-Learning-Enhanced Static Risk-Oriented Security Assessment under Uncertainty

• Published in: 2024 IEEE Power & Energy Society General Meeting (PESGM) pp. 1 - 5
• Main Authors: Masoumi, Amin, Korkali, Mert
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21.07.2024
• Subjects: Accuracy; Complexity theory; Computer architecture; Contingency management; deep learning; generation (resource) adequacy; Power system reliability; Reliability; Scalability; Security; Simulation; Static security assessment; Uncertainty; uncertainty and contingency modeling
• ISSN: 1944-9933
• DOI: 10.1109/PESGM51994.2024.10688480

A power system's ability to supply resources to maintain continuous service to demand, known as generation (resource) adequacy, is an essential constituent of power system planning. Tools and methods to quantify adequacy are fundamental to launching emergency control actions beforehand. As the role of variable renewable energy grows in a decarbonized environment, managing resource diversity over large geographic areas is becoming increasingly critical to maintaining reliability and security, requiring one to revisit the methods to keep up with the changing resource mix and demand. This complexity is exacerbated by prolonged contingencies, inducing system risks. To this end, we propose a fully deep-learning (DL) architecture to overcome the complexity by striking a balance between accuracy and computational speed. The simulation results on the IEEE RTS-GMLC test system strongly favor the DL-based assessment compared to conventional and machine-learning-based counter-parts.