面向拓扑变化的暂态电压稳定智能评估及快速更新策略OA

Aimed at the problem of limited online updating efficiency caused by the complex structure of transient volt⁃age assessment models,a transient stability assessment scheme integrating sample dynamic generation and parameter fine-tuning is proposed in this paper,which is based on an assessment model with spatiotemporal feature extraction ca⁃pability.First,a closed-form continuous-time neural network is employed to extract the hidden features from the time-series data,and neural circuit policies are used to establish a relationship between the assessment model and the grid to⁃pology,thereby enhancing the model's spatiotemporal feature perception and grid topology tracking adaptability.Sec⁃ond,the number of generated samples is dynamically adjusted based on the sample similarity comparison before and af⁃ter thetopology changes,so as to accelerate the construction of the updated sample set and quickly update the model's key parameters using transfer learning.In addition,a time-adaptive assessment method incorporating heterogeneous samples is introduced to strike a balance between the timeliness and accuracy for the online assessment.Finally,case studies on IEEE 39-bus and 145-bus systems validate the effectiveness of the proposed method.