Source–network–load–storage-based coordinated fault-tolerant control for active distribution networks under sensor faults

This paper investigates the problem of fault-tolerant control in source–network–load–storage (SNLS)-based active distribution networks under sensor faults, which pose a significant risk to system stability and reliable operation. To address such issues, a model-free fault estimation method based on deep neural networks (DNNs) is proposed, enabling accurate reconstruction of faulty sensor signals without requiring explicit system models. The recovered measurements are integrated into a distributed control framework to maintain voltage and frequency regulation, along with proportional reactive power sharing among distributed generators (DGs) and distributed energy resources. Compared with existing model-based and rule-driven approaches, the proposed method achieves higher estimation accuracy, improved adaptability, and seamless control integration. Finally, simulation results validate the robustness and effectiveness of the proposed method under both normal and faulty scenarios.