Análisis de vulnerabilidad y resiliencia en sistemas de energía eléctrica ante la ocurrencia de un evento disruptivo deliberado

Abstract

Electrical Power Systems (EPS) face service interruptions caused by deliberate disruptive events, such as cyberattacks, which significantly impact the system's vulnerability and resilience. These attacks compromise the system's ability to recover and restore normal operation, reducing its responsiveness and adaptability, leading to greater economic losses and prolonged impact on critical loads. This Doctoral thesis proposes an optimization model based on the interdiction problem, implementing a Genetic Algorithm (GA) that will identify the system's vulnerable points (interdiction vectors) and prioritize mitigation actions, minimizing vulnerability and maximizing resilience. The vulnerability analysis identified the most susceptible elements to attacks, such as lines and generators. The development of the Genetic Algorithm (GA) extracted the Interdiction Matrix (IM) and the Interdiction Vector (IV), which generate higher costs for the system in the event of a disruptive incident. The most effective strategies of the attacker were identified, targeting lines and generators, as well as the response of the Network Operator (NO). Four scenarios were established, including the base scenario, in which Demand Response (DR) mechanisms and Distributed Generation (DG)/Generation Plants (GP) were applied, aiming to improve the system's resilience. In the transmission system scenario, by applying DR and GP, the system managed to increase the load served by 69%, while applying DR and GP separately resulted in increases of 64% and 47%, respectively. In the distribution system, by applying DR and DG, the system managed to increase the load served by 56%, while applying DR and DG separately resulted in increases of 46% and 44%, respectively. Additionally, a topological reconfiguration strategy of the electrical network was implemented, suggesting alternative configurations after a disruptive event to maximize resilience and minimize operational costs and vulnerability. Finally, metrics were established to quantify and qualify the resilience of the NO's mitigation actions, reducing load loss and its costs, validated through case studies with IEEE test networks. Among the contributions, the most important contribution of this Doctoral thesis is the development of an optimization technique based on Genetic Algorithms (GA) that allows identifying the most damaging attack plans and determining the optimal restoration actions for system elements, such as lines and generators.