Research and application of converter transformer early warning technology based on multi-dimensional analysis

Due to the differences in the operating environment and working conditions of converter transformer equipment, the threshold value set by relevant operation and maintenance specifications has certain limitations in anomaly diagnosis. Existing intelligent diagnosis algorithms have problems such as poor model generalization and failure to consider time series model. From the perspective of engineering application, a multi-dimensional analysis and early warning method for converter transformer is proposed. The three phase transverse comparison and time series trend analysis method are studied for the key parameters of the converter transformer key monitoring disk. The converter transformer leakage oil identification method and oil temperature prediction based on LSTM are proposed to solve the nonlinear and time delay problems in the state parameter prediction. The system is deployed and applied in Suidong Railway Station. The results show that the algorithm model can effectively identify the converter transformer operating state anomalies.