XDT-SG: Explainable AI-Driven Digital Twin Framework for Real-Time Fault Prediction and Self-Healing Smart Grids

Abstract

Smart grid infrastructures have become increasingly complex due to the proliferation of renewable energy sources, Internet of Things (IoT) devices and distributed energy systems. The availability of reliable fault prediction, fault recovery and continuous power supply is great challenge in a dynamic smart grid environment. The current fault management methodologies are mostly reactive in nature with a tendency to delay fault detection, lack of transparency and ineffective recovery processes. In this regard, this paper introduces an Explainable AI-Driven Digital Twin Framework for Real-Time Fault Prediction and Self-Healing Smart Grids (XDT-SG). The proposed framework combines real-time data acquisition, Digital Twin based grid monitoring, Explainable Artificial Intelligence (XAI) based fault prediction and autonomous self-healing capabilities for smart grid management. The Digital Twin model constantly synchronizes the physical and virtual grid environments for proactive monitoring and fault analysis, and the Explainable AI module identifies critical operational parameters that lead to a fault. The framework also intelligently isolates, balances proper load and recovers the power with autonomous recovery operations. The experimental results show the proposed XDT-SG framework can make a better prediction of faults with an accuracy of 98.4%, and it can detect faults with a latency of 0.82 s, recover from faults with a recovery time of 2.14 s and increase the stability of the grid by 97.1% over the conventional methods. The proposed framework provides reliable, transparent and scalable solution for next gen smart grid systems and energy management applications that are intelligent.