Power System Transient Stability Assessment Based on Snapshot Ensemble LSTM Network

Data-driven methods using synchrophasor measurements have a broad application prospect in Transient Stability Assessment (TSA). Most previous studies only focused on predicting whether the power system is stable or not after disturbance, which lacked quantitative analysis of risk transient stability. Therefore, this paper proposes two-stage TSA method based snapshot ensemble long short-term memory (LSTM) network. This can efficiently build an model through single training process, and employ disturbed trajectory as inputs, realize rapid end-to-end TSA. In first stage, dynamic hierarchical assessment carried out classifier, so to screen credible samples step by step. second regressor used predict stability margin undetermined samples, combined with built function quantification angle Furthermore, modifying loss model, it effectively overcomes sample imbalance overlapping. The simulation results show that proposed accurately binary information representing status but also reasonably reflect safety level systems, providing reliable reference for subsequent control.