A Hybrid Model-Driven and Data-Driven Approach for Saturation Correction of Current Transformer

Current transformer (CT) saturation can cause the malfunction of the relay protection devices, which may destroy the stability of the power system under certain circumstances. In this paper, a hybrid physical model-driven and data- driven approach is proposed for the saturation correction of CT. On the premise of constructing the physical model of the undistorted fault current, the data- driven LSTM network is applied to identifying the key parameters of the model. Benefitting from the excellent performance of the LSTM network in processing timing series signals, this method has the advantages of flexible valid data window and low sampling rate, etc. In addition, it is worth mentioning that the attention mechanism is introduced and provides a visual explanation for the results. Finally, abundant simulation results verify the effectiveness of the method proposed in this paper.