Date of Award

2014

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical, Computer and Biomedical Engineering

First Advisor

Yan (Lindsay) Sun

Abstract

Electrical grids have been developed over a century, which are considered as one of the most important infrastructures on the earth. In the past decade, the emergence of the Smart Grid, referred to the next generation of power grid, makes existing systems more complicated and vulnerable. Cyber-physical attacks against existing systems and future smart grids have drawn increasing attention, because such attacks could trigger large-scale cascading failures and result in major blackouts.

In the traditional power society, contingencies are widely considered as the causes that result in power outages. The contingency analysis is the predominant method to investigate the vulnerability of power grids. With the increasing malicious attacks against power transmission systems, however, studying the grid's security and reliability only from the contingency analysis perspective has apparent limitations. First, contingencies happen randomly and unintentionally; malicious attacks are mostly intentional. Second, it is rare that multiple contingencies happen simultaneously. Malicious attacks, however, can likely occur on a few, even more, the power grid components.

In this dissertation, the security and reliability of power grids is investigated. Briey speaking, the attackers identify a few components in the grid as targets (e.g., substations, transmission lines, or both). Then, the attackers take down these targets by either physical sabotages or cyber intrusions, hoping that the initial failures can trigger large-scale cascading failures. The goal of the attackers is to find a group of targets, attacking on which can yield large damage to the power grid. In particular, this dissertation investigates the attacks against the power system from the following aspects.

It is a nature question that why attacking a few, even one, critical components can severely weaken the system. In manuscript 1 (i.e., chapter 2), the cascading process is visualized to help people under such complicated phenomena, as well as discovering different types of failure propagation.

Attackers might only know the topological connection of the power grid, e.g., the topology. In manuscript 2 (i.e., chapter 3), a topology-based cascading model is adopted to study cascading failures. The metric load distribution vector (LDV) and LDV-based attack strategy are proposed and investigated.

Attackers can possibly know some general information of the power grid, e.g., the topology, types of substations and length of transmission lines. In manuscript 3 (i.e., chapter 4), the extended topological model is used to mimic cascading failures. A novel metric, called the risk graph, is proposed to reveal the hidden relationship among critical substations/transmission lines. In addition, the risk-graph based attack strategies are developed regarding substations and transmission lines, respectively.

Attacks can occur on substations and transmission lines simultaneously. In manuscript 4 (i.e., chapter 5), both the vulnerability analysis and the attacks are investigated from the joint substation-transmission line perspective.

Attacks can be conducted not only synchronously but sequentially. In manuscript 5 (i.e., chapter 6), the sequential attack is introduced; the metric sequential attack graph (SAG) is constructed; the SAG-based sequential attack strategy is developed and evaluated.

Share

COinS