Cascading Failure Path Prediction based on Association Rules in Cyber-Physical Active Distribution Networks

Cascading failures may lead to large scale outages, which brings about significant economic losses and serious social impacts. It is very important to predict cross-domain cascading failures paths for identification of weak nodes, which contributes to the control policies for preventing cascading failures and blocking their propagation between cyber domain and physical domain in cyber-physical active distribution networks. This paper proposes an algorithm based on the Frequent-Patterns-Growth (FP-Growth) to predict cascading failure paths, which predicts the potential failure node set by analyzing a large number of simulation datum and mining the hidden association relationship among datum. To demonstrate the effectiveness of the proposed cascading failure path prediction approach, an empirical study on a cyber-physical active distribution network, named CEPRI-CPS from Electric Power Research Institute of China, is performed, and the result shows the robustness of cyber-physical active distribution networks can be improved with prediction approach in this paper.