Fault Diagnosis for Distributed Grids and Frontier Exploration of Machine Learning Methods

Fault diagnosis in distributed power grids is very important for keeping the grid safe and reliable. As distributed generation grows, traditional centralized power grids are changing. They are becoming more decentralized and complex. This creates new challenges for traditional fault diagnosis methods. Distributed power grids have bi-directional power flow, different distributed energy sources, and intermittent generation. These features make fault diagnosis harder. In this review, the characteristics of distributed grids and common fault types are introduced. Next, this paper categorizes and summarizes the mainstream distributed grid fault diagnosis methods, primarily concentrating on the methods of signal processing-based, modeldriven, and machine learning-based. Finally, the review investigates the key challenges and future research directions and trends of distributed grid fault diagnosis.