Modeling and Stability Analysis of Self-Evolving Models with Small Sample Data Based on Grid-Forming Converter

Although the traditional stability theory based on synchronous motor characteristics has been improved and validated for over a century, it has limitations in dealing with the rapid changes brought by renewable energy sources and high-power electronic devices in modern power systems. Research into supporting technologies has become an indispensable element in the evolution of the new power system, including control algorithms designed for power electronic devices, energy storage options, and advanced intelligent control technologies. This paper proposes a multi-task learning system for small sample data learning and provides an algorithm for processing and classifying oscillation sample data, optimizing the data sample. In addition, the GFM converter impedance angle model and temporal model are established, and the conditions for stable operation of the GFM converter are analyzed. Finally, the conditions for stable operation of the system are verified thoverrough simulation.