Transformers are core equipment in the power distribution network of power systems. Accurately and rapidly acquiring the dynamic temperature of winding hot-spots is crucial for ensuring the safe and stable operation of transformers and the reliable power supply of the power grid. However

the internal structure of transformers is complex

and the characteristic parameters of insulating oil are susceptible to impurities

temperature

and other factors

making it difficult to accurately monitor the dynamic temperature of winding hotspots. Therefore

this paper takes the S11-M-630/10 type oil-immersed distribution transformer as the research object to study the dynamic inversion method for the winding hot-spot temperature of oil-immersed transformers. First

the streamline method is adopted to select characteristic points on the transformer shell. Second

the LassoNet algorithm is combined to determine the optimal set of characteristic temperature measurement points

and a multi-parameter-based LassoNet-ISSA-BP dynamic inversion model for winding hot-spot temperature is proposed. The accuracy of the model is validated under eight typical experimental operating conditions. The results show that the maximum error between the inversion value and the measured value of the proposed method is 2.21 ℃

and the average error is lower than 0.79 ℃

which is significantly better than that of the IEEE standard calculation. This paper provides a feasible approach for the rapid

accurate

and convenient acquisition of the dynamic winding hot-spot temperature of transformers and offers a technical support for on-site evaluation of transformer temperature rise performance using small-current test data.