Intelligence Augmented Transient Stability Discrimination via Adaptive Critical Threshold Modification of sBTTC

The transient stability discrimination (TSD) forms the basis for the state restoration and secure operation of power systems after faults occur. Mechanistic criteria in existing research often suffer from limited applicability and timing constraints while artificial intelligence-dominated TSD lacks interpretability and real-time reliability assessment. In the regard, this paper delivers a novel TSD method based on the intelligence augmentation of the mechanistic criterion. Firstly, targeting the minimal value prediction of simplified branch transient transmission capacity (sBTTC), a deviation variance upper bound estimation algorithm is put forward to measure the expected error distribution of prediction results. Then, in combination with the probabilistic distribution function and transient instability probability, a risk penalty function which balances the timing requirement and accuracy is introduced with the extreme value condition proven. Finally, by continuous comparison between calculated value and the optimal threshold of sBTTC deduced by the Newton method, the TSD is performed. CSEE-RAS and actual large-scale power system cases verify the validity, interpretability, and precision of the proposed intelligence-augmented criterion.