A Joint Data-Physics-Driven Approach to Voltage Sag Source Tracing and Inversion Estimation

With the global energy transition and the development of smart manufacturing, modern power systems are characterized by a high penetration of renewable energy and power electronic devices, leading to a declining resilience against voltage sags. Accurate post-event source identification and event inversion estimation form the foundation for managing and mitigating voltage sag issues, which is of great significance for enhancing high-quality power supply in the grid. This paper proposes a method for voltage sag source identification and inversion estimation based on disturbance energy and pattern matching, aiming to achieve precise localization and full-process inversion estimation of voltage sag events, thereby providing key technical support for refined grid management and high-quality power supply. First, by analyzing the changes in energy flow direction at each monitoring point before and after a fault, upstream and downstream identification of the sag source is achieved. Combined with judgment results from multiple monitoring points, the exact location of the voltage sag is accurately determined, overcoming the challenge of insufficient source identification accuracy caused by multi-directional power flow from renewable energy sources in active looped systems. Second, pattern matching technology is introduced to construct a voltage sag event inversion estimation model. Using the Monte Carlo simulation method, a voltage sag source identification pattern library is generated. The measured data from limited monitoring points are intelligently matched with the pre-generated pattern library based on characteristics, enabling post-event inversion of voltage sag events and determining the voltage sag severity at unmonitored nodes. Finally, the IEEE 30-bus system is used as a case study to simulate and verify the effectiveness and accuracy of the proposed method in precise voltage sag source localization and inversion estimation.