Date of Award

August 2015

Degree Type


Degree Name

Master of Science



First Advisor

Lingfeng Wang

Committee Members

David C. Yu, Yue Liu


Cascading Failure, Critical Components Identification, Cyber-physical Security, Smart Grid


As a revolutionary change to the traditional power grid, the smart grid is expected to introduce a myriad of noteworthy benefits by integrating the advanced information and communication technologies in terms of system costs, reliability, environmental impacts, operational flexibility, etc. However, the wider deployment of cyber networks in the power grid will bring about important issues on power system cyber security. Meanwhile, the power grid is becoming more vulnerable to various physical attacks due to vandalism and probable terrorist attacks. In an envisioned smart grid environment, attackers have more entry points to various parts of the power grid for launching a well-planned and highly destructive attack in a coordinated manner. Thus, it is important to address the smart grid cyber-physical security issues in order to strengthen the robustness and resiliency of the smart grid in the face of various adverse events. One key step of this research topic is to efficiently identify the vulnerable parts of the smart grid.

In this thesis, from the perspective of smart grid cyber-physical security, three critical component combination identification methods are proposed to reveal the potential vulnerability of the smart grid. First, two performance indices based critical component combination recognition methods are proposed for more effectively identifying the critical component combinations in the multi-component attack scenarios. The optimal selection of critical components is determined according to the criticality of the components, which can be modeled by various performance indices. Further, the space-pruning based enumerative search strategy is investigated to comprehensively and effectively identify critical combinations of multiple same or different types of components. The pruned search space is generated based on the criticality of potential target component which is obtained from low-order enumeration data. Specifically, the combinatorial line-generator attack strategy is investigated by exploring the strategy for attacking multiple different types of components. Finally, an effective, novel approach is proposed for identifying critical component combinations, which is termed search space conversion and reduction strategy based intelligent search method (SCRIS). The conversion and reduction of the search space is achieved based on the criticality of the components which is obtained from an efficient sampling method. The classic intelligent search algorithm, Particle Swarm Optimization (PSO), is improved and deployed for more effectively identifying critical component combinations.

MATLAB is used as the simulation platform in this study. The IEEE 30, 39, 118 and Polish 2383-bus systems are adopted for verifying the effectiveness of the proposed attack strategies. According to the simulation results, the proposed attack strategies turn out to be effective and computationally efficient. This thesis can provide some useful insight into vulnerability identification in a smart grid environment, and defensive strategies can be developed in view of this work to prevent malicious coordinated multi-component attacks which may initiate cascading failures in a cyber-physical environment.