Transient State Estimation for Power System Based on Deep Transfer Learning
A method for transient state estimation in power systems based on deep transfer learning is proposed to accurately track transient state in real-time,which is typically challenging owing to the limited availability of fault sample data.Initially,the twin data representing the actual power system operation are generated by utilizing digital twin technology,thereby providing substantial sample data sources for transient state estimation.Subsequently,the twin datasets are partitioned into source domain and target domain datasets,and a base model is developed for state estimation in the source domain based on steady-state power system data.Finally,by applying deep transfer learning,the base model is fine-tuned using small-sample transient data in the target domain,resulting in a state-estimation model specifically adapted for transient conditions and enhancing the universality of the estimator.Simulations demonstrate that the proposed method exhibits a higher estimation accuracy and computational efficiency than that of deep neural networks without transfer learning,particularly during power system failures.
power system faultdigital twintransient state estimationdeep transfer learningsmall sample
国家科技期刊平台
NSTL
万方数据
