Advanced Probabilistic Transient Stability Assessment for Operational Planning: A Physics-Informed Graphical Learning Approach

Existing probabilistic transient stability assessment (PTSA) methods mainly provide an overall estimation of the probabilistic transient stability index (TSI) but ignore the temporal characteristics of each individual synchronous generators. In addition, conventional surrogate model-based PTSA trains a model for each individual trip, ignoring uncertainties of random trips. To address the above challenges, this paper develops a physics-informed graphical learning approach for PTSA to predict the post-fault rotor angle trajectories (based on the pre-fault system state and trip location) and deal with multiple trips. The statistical analysis is designed for the developed advanced PTSA to achieve two objectives. First, it provides an overall estimation of the probabilistic TSI by using the maximum rotor angle difference. Second, to further improve the situational awareness of operators, it visualizes the temporal information of the probabilistic transient stability. Three-sigma rule is used to analyze the trajectory of TSI (mean, upper bound, and lower bound). To visualize the probabilistic TSI at any assessment time point, the TSI PDF at the corresponding time point is calculated. Comparison experiments are performed on the IEEE-39 Bus System and IEEE-118 Bus System to verify the efficiency and accuracy of the proposed method.