Research

The reliable operation of modern power systems depends on real-time situational awareness, requiring the deployment of a wide range of sensors, including Phasor Measurement Units (PMUs), Smart Meters, and Distributed Fiber-Optic Sensors (DFOS). However, these technologies involve high costs and demand optimized placement and efficient data transmission for effective grid monitoring. Addressing these challenges requires determining the optimal placement of PMUs and PDCs in large-scale transmission grids, as well as identifying strategies for deploying next-generation smart meters in distribution networks. Additionally, integrating fiber-optic sensing with traditional monitoring techniques and designing real-time, low-latency communication strategies are crucial for seamless data transmission across the grid.

To overcome these challenges, we developed optimal PMU and PDC placement strategies for large-scale power grids and proposed a hybrid sensor deployment framework integrating PMUs and DFOS. Additionally, we explored second-generation smart meter placement and management strategies for enhanced grid-edge monitoring. Finally, we designed efficient data routing strategies to enable real-time, resilient data transfer, ensuring reliable communication for grid monitoring applications.

With millions of sensors deployed across the power grid, system operators receive a massive amount of raw data. However, translating this data into actionable insights is essential for real-time decision-making. Techniques such as observability analysis, state estimation, anomaly detection, and forecasting play a crucial role in establishing grid situational awareness, particularly in low-observable distribution networks where real-time sensor data may be limited, delayed, or unreliable. Additionally, synchronizing data from heterogeneous sensors and ensuring accurate state estimation remain key challenges.

To address these challenges, we developed AI-assisted state estimation techniques to provide state variables in high-resolution. Additionally, we built fault localization models using deep learning and hybrid sensor data, including DFOS and uPMUs, to improve localization accuracy. Furthermore, we proposed deep learning-based load and voltage estimation frameworks to enable real-time grid situational awareness.

Cyber-physical power systems are vulnerable to physical failures (e.g., equipment failures) and cyber threats (e.g., data manipulation, false data injection attacks). To ensure reliability, the grid must be self-healing, adaptive, and resilient in minimizing disruptions. Achieving resilience requires strategies that address both cyber and physical failures, along with optimal operation of controlled equipment, such as DERs and switches, to maximize electricity supply during disruptions. Additionally, cyber-physical coordinated anomaly detection methods are essential for real-time cyberattack detection and mitigation.

To address these challenges, we proposed cyber-physical reconfiguration strategies to maximize electricity supply during disruptions. We also developed staggered sensor-data-driven volt-var control strategies for distribution systems including the secondary networks, facilitating grid operation. Furthermore, we designed large language model-based reconfiguration strategies that effectively minimize system loss, providing quality electricity.