Abstract
This paper introduces a novel approach for managing fault propagation in interconnected energy networks comprising electric, gas, and heating systems. As energy infrastructures become increasingly integrated, the risk of cascading failures across these networks grows, making it critical to develop robust models for predicting and mitigating fault propagation. To tackle the complexity of fault propagation in interconnected energy systems, we develop a novel AI-based management architecture that couples adversarial learning mechanisms with graph-structured predictive models. Specifically, a generative network is employed to synthesize plausible fault evolution patterns from historical records, while a graph-based neural architecture captures the spatiotemporal correlations among the subsystems. Furthermore, a robust optimization scheme under distributional uncertainty is incorporated to devise adaptive recovery strategies, enhancing the resilience and reliability of system restoration processes. The proposed model is tested using a synthetic case study based on the IEEE 123-bus electric network, the Belgian gas transmission system, and a standard heating network. The results demonstrate the effectiveness of the model in accurately predicting fault propagation and optimizing recovery strategies, significantly reducing recovery time and minimizing the impact of cascading failures. The primary innovations presented in this study are the development of an integrated fault propagation framework spanning multiple energy networks, the pioneering use of adversarial and graph-based learning techniques for fault trajectory prediction, and the incorporation of distributionally robust optimization to strengthen recovery planning. Collectively, these advancements contribute to a deeper understanding of fault dynamics in interconnected infrastructures and propose a scalable pathway for enhancing system resilience in modern energy networks.