Coordinated Source-Grid fault current limiting strategy in HVDC grids based on fault severity and quantitative limitation demand

• Machine-learning predicts fault current in 1 ms for severity grading. • HB-MMC dual-index control integrates peak and energy limits for adaptive limiting. • Four-level severity scheme optimizes activation timing of source–grid coordination. Reliable fault current limitation is critical in high-voltage direct current (HVDC) grids due to the fast dynamics and low fault tolerance of power electronic equipment, as well as the high cost of DC circuit breakers (DCCBs). To address these challenges, this paper proposes a coordinated source-grid fault current limiting strategy based on fault severity assessment and quantitative measurement of current limitation demand. Firstly, an artificial neural network (ANN)-based model is developed to predict 1-ms fault currents in real time, enabling rapid classification of fault severity within 1 ms. Secondly, a dual-metric approach is introduced to quantify the current limiting requirements from both the current amplitude and energy perspectives, building a modulation coefficient controller for adaptive source-side suppression. Thirdly, an adaptive grid-side fault current limiter (FCL) operation logic is formulated, where the engagement timing and coordination with the source side are optimized according to the assessed fault severity. Simulation and experimental results demonstrate that the proposed strategy reduces DCCB breaking current by over 44 % and metal-oxide arrester (MOA) energy absorption by more than 75 %, compared with conventional approaches. The method enhances system protection, improves coordination between devices, and is particularly suited to large-scale renewable-integrated HVDC networks.