Hierarchical Time-Series Assessment and Control for Transient Stability Enhancement in Islanded Microgrids

The grid-forming converters would integrate battery energy storage systems (BESSs) in islanded microgrids for smoothing out the uncertain fluctuations of renewable energy resources. However, gird-forming converters would pose transient instability risks under large disturbances, which could require fast and accurate stability assessment methods in such challenging and uncertain cases. In this work, a hierarchical time-series assessment and control (HTSAC) framework is proposed for assessing the transient stability of grid-forming converters in islanded microgrids. The proposed HTSAC framework offers a gated recurrent unit (GRU) neural network alternative for an intuitive and accurate trajectory prediction in early post-fault stages. Subsequently, an emergency ride-through control (ERC) strategy is proposed which leverages the proposed neural network approach for enhancing the prediction results in real-time assessments of microgrid transient stability. The initial input horizon of GRU is optimized to avoid intense trial-and-error design burdens incurred in conventional data-driven assessment methods. Simulation and experimental results are presented to validate the effectiveness of the proposed HTSAC on an islanded microgrid in south China. The results also point out that the GRU of the prediction layer with a quantile loss function would ensure a timely ERC in the proposed HTSAC approach to renewable energy-based converter operations.