Virtual sensing of transformer winding hot spot temperature based on transformer digital twin model

As transformer oil-paper insulation deterioration is significantly affected by the winding hot spot temperature (HST), it is of great significance to realize the accurate prediction of the HST. Based on the multi-physical field simulation analysis model of transformer and support vector regression artificial intelligence algorithm, the digital twin model of a S13-M-400 kV A/10kV transformer is constructed to realize the virtual sensing of HST. The coupled electromagnetic thermal fluid field analysis is realized by the method of indirect coupling, and the established multi-physical field simulation analysis model is used to calculate the working conditions of training samples and test samples. A good results are obtained in the test samples, with an average relative percentage error of 2.18% and a maximum virtual sensing error of 3.17 °C. The established virtual sensing model is applied in the HST detection of a S13-M-400 kV A/10kV transformer in service, which can help continuous monitoring winding HST, provide references for the formulation of transformer operation strategies and evaluation of transformer insulation life.