When a short-circuit fault occurs along the transmission line of a modular multilevel converter high-voltage direct-current (MMC-HVDC) grid, the sub-module capacitors discharge, causing the fault current to rapidly rise, posing a threat to the safe operation of the system. Therefore, this paper proposes a novel fault location method based on the Gram Angle Difference Field (GADF). The column corresponding to the maximum differential value in a sliding window is used to identify the fault moment and locate faults in MMC-HVDC transmission lines. In order to effectively distinguish normal fluctuations from fault mutations and avoid false alarms, a dynamic threshold is set based on the statistical characteristics of normal data. This method utilizes the unique feature extraction capability of the GADF matrix, the adaptive mechanism of the dynamic threshold, and the stability of line-mode voltage to achieve fast and accurate fault location. Finally, this method is validated using a simulation model. The results show that the proposed method can accurately locate faults in different conditions.

Modular multilevel converter high voltage direct current (MMC‐HVDC) grids are of great significance in solving problems such as long‐distance power transmission and large‐scale renewable energy grid connection. However, their complex structure and unique operating characteristics pose challenges to transmission line protection, particularly under high‐resistance faults and noise interference. Existing methods based on single physical criteria or end‐to‐end data‐driven approaches have inherent limitations in sensitivity, reliability and interpretability. To address these issues, this paper proposes a novel collaborative protection method integrating physical mechanism guidance and attention‐based deep learning. First, a physical criterion is constructed based on voltage differential energy to achieve rapid and reliable fault section identification. Second, by introducing a gram angle difference field, the one‐dimensional voltage time‐series signal is converted into a two‐dimensional feature image that retains time correlation, effectively enhancing the expressive power of fault features. Finally, a multi‐attention convolutional neural network model is designed, which uses adaptive focusing through channel and spatial attention mechanisms to achieve accurate extraction and classification of key fault features. Simulation results show that the proposed method can achieve high‐precision fault identification of the MMC‐HVDC system under various operating conditions and has strong anti‐noise ability and high fault resistance.