Online time series monitoring method of transformer based on seq2seq model

Abstract

Abstract To monitor the operating conditions of transformers in real time and improve the monitoring accuracy of three-phase voltage, load current and power frequency of transformers, an online time series monitoring method of transformers based on the seq2seq model is proposed. The impedance, excitation current, secondary side voltage and excitation branch of the transformer are reduced to the primary side value, and the regulation coefficient of the transformer under no-load operation is calculated. By obtaining the short circuit impedance of the transformer, the short circuit impedance of the transformer is analysed. Using the coupling inductance model structure of single-phase double-winding and single-phase three-winding transformers, the parameter matrix of a single-phase double-winding transformer is established. According to the corresponding relationship between port electrical data and online monitoring information, the transformer coupling inductance model is constructed. The online time series monitoring process of the transformer is designed by using the seq2seq model, and the online time series monitoring of the transformer is realized. The experimental results show that when monitoring the three-phase voltage, load current and power frequency of the transformer, the monitoring errors are controlled below 0.5, 5 and 0.5%, respectively, which greatly improves the accuracy of online time series monitoring of the transformer.