Power Market Cybersecurity and Profit-targeting Cyberattacks

The COVID-19 pandemic has forced many companies and business to operate through remote platforms, which has made everyday life and everyone more digitally connected than ever before. The cybersecurity has become a bigger priority in all aspects of life. A few real-world cases have demonstrated the current capability of cyberattacks as in [1], [2], and [3]. These cases invalidate the traditional belief that cyberattacks are unable to penetrate real-world industrial systems. Beyond the physical damage, some attackers target financial arbitrage advantages brought by false data injection attacks (FDIAs) [4]. Malicious breaches into power market operations could induce catastrophic consequences on fair financial settlements and reliable transmission services. In this dissertation, an in-depth study is conducted to investigate power market cybersecurity and profit-targeting cyberattacks. In the first work, we demonstrate the importance of market-level behavior in defending cyberattacks and designing cyberattacks. A market-level defense analysis is developed to help operators identify cyberattacks, and an LMP-disguising attack strategy is developed to disguise the abnormal LMPs, which can bypass both the bad data detection and market-level detection. In the second work, we propose a comprehensive CVA model for delivering a detailed analysis of four aspects of vulnerability: highly probable cyberattack targets, devastating attack targets, risky load levels, and mitigation ability under different degrees of defense. In the third work, we identify that revenue adequacy, a fundamental power market operation criterion, has not been analyzed under the context of cybersecurity, and we explore the impact of FDIAs targeting real-time (RT) market operations on ISO revenue adequacy analytically and numerically. In the last work, we extend the power system cybersecurity analysis to multi-energy system (MES) framework. An optimally coordinated (OC-FDIA) targeting MES is proposed. Then, we show that the OC-FDIA cause much more severe damages than single-system FDIA and uncoordinated FDIAs. Further, an effective countermeasure is developed against the proposed OCFDIA based on deep learning technique (DL).