Interaction Models for Analysis and Mitigation of Cascading Failures

In this chapter, a thorough discussion is provided for the cascading failure interaction analysis approach, based on either simulated cascading outage sequences or utility outage data, considering either only line outages or the interdependency between multiple types of outages. The component failure interactions are estimated using an expectation maximization (EM) algorithm, and interaction networks are further built to reveal the overall outage propagation patterns. Critical components can be identified utilizing the interaction networks, with or without considering spatial propagation. Highly probabilistic interaction models using the estimated interactions can help efficiently simulate cascading sequences that match the statistics of the original data, greatly augment the data samples, and enable an efficient study of mitigation strategies. Cascading failure mitigation strategies can also be designed based on the identified critical components to reduce the risk of cascading.