| dc.description.abstract | Electric power grids are critical infrastructure that enhance the security, economy, and productivity of nations. In modern societies, power grids are continually evolving into smarter grids which incorporate automation to improve reliability using information and communication technologies, forming the cyber layer. The cyber layer interacts with the physical power grid by acquiring measurements from, and sending control commands, to field devices, forming an interconnection between the cyber and physical layers, referred to as the cyber-physical power grid.
Although these layers are essential to grid operation, they are vulnerable to a myriad of threats which can evolve to high impact disruptive events, resulting in catastrophic failures and losses such as widespread blackouts, loss of critical services enabled by electricity, and loss of lives and property. These events often cripple economic operations, disrupt societies, and threaten national security. Hence, strengthening the power grid against these threats has easily become a top priority, achieved by improving the resilience of the power grid.
This dissertation presents risk reduction against cyber and physical threats via a resilience-oriented perspective, intended to build into the vision of next generation energy management. The work presented in this dissertation focuses on common threats that have begun to more frequently affect the reliability and resilient operations of power systems, some leading to bankruptcy and strained customer relations for several utilities. Hence, this dissertation answers the question: How Can We Proactively Reduce Critical Power Infrastructure Risk to High Impact Cyber and Physical Threats, Automating the Risk Reduction Process, with Resilience at the Forefront?
Toward this objective, this dissertation first presents an approach based on the axiomatic design process to enable the standardization of power system resilience, an issue that has been elusive to the power system resilience community in the past decade, which elucidates the studies herein. Then, the threats which highly impact the resilience of the power grid as a cyber-physical system are introduced, since the power grid is threatened by adversaries from both the cyber and physical domains. In the cyber layer, the dissertation presents risk reduction to threats of adversary intrusion and ensuing false data injection attacks using different techniques centered around graph-based modeling, where we propose a proactive framework to reduce the impact of adversary intrusion on the system, and develop a detector for stealth attacks which evades conventional power system detectors, respectively. Further on the cyber layer, the dissertation proposes techniques and implements modeling via emulation which achieve automation in the crucial provision of sandbox environments for the risk evaluations of critical infrastructure. These aid to improve system resilience via use cases such as cyber deception where redundancy against adversaries is provided to power system networks. In the physical layer, the dissertation focuses on the frequent and high impact threat of wildfires. The risk minimization builds on accurately modeling wildfire threats in a proposed novel technique that is designed to be efficient for the bulk power grid as opposed to conventional methods in which power systems adapt techniques better suited for wildlands. The proposed technique uses spatio-temporal and data-driven deep learning methods, which can effectively reduce power system risk from endogenous wildfires caused by the power grid, and exogenous wildfire from external sources. Beyond risk assessment and minimization, the dissertation proceeds to present the first-of-its-kind resilience-comprehensive design and development of a self-sufficient low-cost wildfire mitigation model which automates the risk reduction process towards mitigating wildfires in grid operation through all phases in which the power system lies before, during, and after a wildfire threat or event. | |
