基于多对抗迁移学习的暂态稳定评估模型OA

Transfer learning has been introduced to power system transient stability assessment(TSA)to expand scenario coverage.However,when transferring the classification boundary knowledge from known faults to potential fault assessments,existing methods often exhibit low accuracy for critical samples in the target domain.To address this,this paper proposes a multi-adversarial transfer learning model with multi-domain discriminators.By incorpo-rating fault severity indices as prior knowledge,fault samples are subdivided into four classes.Multiple domain dis-criminators then align these four sample categories between source and target domains.Through a multi-adversarial adaptation framework,granular alignment of sample distribution is achieved.This approach significantly improves the assessment accuracy for critical samples in the target domain while enhancing the model's positive transfer capa-bility.Simulation results on the IEEE 39-bus system and a regional power grid validate the effectiveness of the pro-posed method.