Resilient operating strategies against weather-related and cybersecurity-related events

Electric power systems are considered as critical infrastructures of modern societies. To support sustainable developments of modern societies, electric power systems are being constructed with the characteristics of better economy and higher reliability via many ways, e.g., power system interconnection and cyber-based monitoring/controls. Power system interconnection results in wide geographical coverage, which makes power systems exposed to external weather-related events. In addition, cyber-based systems for monitoring/controls make power systems more vulnerable to potential cyber intrusions. Conventional strategies, due to little consideration of weather-related and cybersecurity-related events, fail to make appropriate responses to these events. To deal with potential weather-related and cybersecurity-related events, power systems are suggested to have the characteristic of resilience. The key understanding of resilience is to proactively enhance robustness prior to an event, real-timely mitigate damages during an event, and rapidly restore systems after an event. Though the characteristics of resilient power systems have been well recognized, specific resilient strategies are still in the infancy. The contributions of this thesis are to establish sequential strategies against weather-related and cybersecurity-related events to construct a resilient power system. First, prior to weather-related events, a dynamic coordinated condition-based maintenance strategy, considering influences of extreme weather events, is proposed. The deterioration of each component and the influences of weather events are modeled as Markov processes. A backward induction with a search space reduction method is employed to improve computational efficiency while still maintaining good accuracy. Second, during weather-related events, two points are focused on. The first point is to develop an extended Kalman filter with equality constraints for multi-area dynamic state estimation to help system operators achieve system states. In the proposed method, a corrective strategy is proposed to estimate a corrective internal voltage (CIV) and a corrective rotor angle (CRA) of each generator. The CIV and the CRA are then used to establish equality constraints for updating generators’ dynamic states. The second point is to propose a Markov-based model for sequential generation redispatch during weather-related events. The system topology, at each decision epoch, is considered as a Markov state. For each state, a recursive model, including a current cost and a future cost, is established, considering operation constraints. By using the linear scalarization method, the recursive model is transformed into a linear programming model. Third, during cybersecurity-based events, three points are focused on. The first point is to propose a cyber inference system against alter-and-hide attacks. The second point is to propose a dynamic game-based model to determine preventive strategies for mitigating the risk of power outages when a system is under cyberattacks but before unauthorized operations can be conducted by intruders. The third point is to propose a risk mitigation model with a partially observed Markov decision process, when a PMU-based grid is under cyberattacks.published_or_final_versionElectrical and Electronic EngineeringDoctoralDoctor of Philosoph