Data-Physics Integrated Method for Cascading Faults Identification

As the proportion of renewable energy sources and High Voltage Direct Current (HVDC) integration into power systems increases, cascading faults have garnered more attention due to their potential risks and severe consequences. However, due to the vast amount of data and rapid changes in actual power grids, it is challenging to quickly identify the type and location of occurring faults. To address this issue, this paper proposes a data-physical fusion method for the identifying and tracing typical power system faults. The proposed method combines the mechanism of electromagnetism during transient stability and the capability of feature extraction of Convolutional Neural Networks (CNN). Case studies have shown that the proposed method can identify the typical faults with the accuracy of 98.1%.