Transient Stability Assessment of Power Systems Based on CK-means-CNN

Under the background of the energy transition, the transient stability assessment of power systems faces new challenges. Traditional methods for assessing the transient stability of power systems have limitations in complex systems. Although artificial intelligence technologies have been widely applied in this field, issues such as data imbalance, model interpretability, and economic efficiency restrict their development. This paper proposes an assessment model that integrates the CK-means algorithm and a convolutional neural network (CNN). Using positive-sequence voltage and positive-sequence angle as inputs, the model effectively alleviates data imbalance through CK-means and oversampling, and utilizes the powerful feature learning ability of the CNN model to deeply mine the key feature information in transient data. Verified by the simulation results of case studies, the proposed model can effectively identify unstable samples, significantly reduce the risk of misjudgment, and demonstrates excellent comprehensive performance.