Congrats to Fayha for successfully defending her Ph.D. thesis.
The continuous and accelerated digitalization of industries and technologies has made most of our daily activities obtrusively depend on electricity. Consequently, reliable power system operation became the cornerstone of economic sustainability and technological development. Unfortunately, the grown dependency of modern power infrastructure on Information and Communication Technology (ICT) has increased the risks of cyber-attacks. According to the most recent statistics, the electrical power sector is one of the significant fields in the number of cyber-attacks per year. The most devious types of cyber-attacks target the power system state estimation. Realtime state estimation aims to filter out the noise of measurements to obtain accurate estimates of the system state. The state estimation plays an essential role at the core of Energy Management System (EMS). Since most of the high-level applications rely on the accuracy of the system stateâs estimates, cyber-attacks targeting the state estimation in an electrical power system can jeopardize the integrity and economic operation of the entire system. In particular, False Data Injection Attacks (FDIAs) and GPS Spoofing Attacks (GSAs) are the two types of attacks with the most severe consequences. Under certain conditions, both attacks can even evade the conventional Bad Data Detection (BDD) techniques. Combined, they can critically impact the ordinary operation of an electrical power system. The majority of the methods studied in the literature assume large errors introduced during the attacks, which often oversimplifies the solution models, therefore influencing their detection accuracy and restricting their mitigation efficacy. In practice, subtle inaccuracies in the system state estimates can yet carry devastating consequences. Nowadays, variable Renewable Energy Sources (RES) are becoming more and more common in modern hybrid power systems. The volatile nature of RES introduces frequent power and voltage fluctuations which increases the spatiotemporal complexity of the system states. Consequently, cyber-attacks that leverage relatively small magnitudes can bypass traditional detection mechanisms. In contrast, deep learning techniques have proven to detect the slightest anomalies in a time series of observations. Furthermore, their fast response makes them the best fit for such real-time applications. This research study aims to investigate the feasibility of employing advanced deep learning architectures to secure the integrity of power system state estimation through anomaly detection and real-time corrections.