Study of robustness of cyber-coupled power systems from a complex network perspective

Abstract

This thesis studies the robustness of smart grids from a complex network perspective. A power system can be modeled as a network consisting of nodes representing power substations and links representing power transmission lines. To study the cascading failure in power systems, we develop a network-based model combining a circuit-based power flow model with a stochastic model. Considering the effect of cyber coupling, a smart grid can be modeled as a cyber-coupled power system in which a power network is connected to a cyber network. To produce the propagation profile of the cascading failure caused by cyber attack, we further introduce a model by considering power overloading, contagion, and interdependence between power and cyber networks. The main objective of this thesis is to enhance the robustness of standalone power systems and power systems that are coupled with cyber networks by taking network-based approaches. First, by examining the propagation profile of the failure cascade of power systems, we define the onset time as the time after which the propagation rate of the cascading failure increases rapidly. Based on the onset time and the scale of the failed grid in a cascading failure event, each component in a power network can be categorized into three types, corresponding to three levels of severity of the failed grid upon the initial failure of that component. Moreover, we propose a decision-tree-based learning model to enhance the robustness of power networks. The resulting decision tree identifies three network features in a power network, including average shortest path length, average clustering coefficient, and average effective resistance (distance) to the nearest generator, which are highly correlated with the network robustness and can effectively contribute to robustness enhancement of the power network. Then, we aim to find an effective interpretation of the interdependence between power and cyber networks in studying the cascading failure in cyber-coupled power systems. We consider the interaction between two processes,one aiming to attack and the other aiming to defend the components in the power network. Through evaluating the effectiveness of different attack and defense strategies by examining the actual propagation process of cascading failure events, it has been found that the tit-for-tat defense strategy, in which the defender adopts the same strategy as the attacker, is the preferred defense strategy. Moreover, allocating defense strength in terms of capacity-based distribution can most effectively suppress cascading failure. Finally, we introduce a parameter, called relative coupling correlation coefficient, to quantify the coupling pattern of a cyber-coupled power system. In modeling coupled systems, coupling patterns, which are determined by some node criticality metrics, can describe how power and cyber nodes are connected. Simulation results show that a coupled system of lower relative coupling correlation coefficient has better robustness. Moreover, when optimizing the coupling pattern for robustness improvement, the adoption of node capability and node degree as node criticality metrics for power and cyber networks, respectively, would result in a much more robust network compared to the adoption of other node criticality metrics for robustness enhancement.