Abstract: Transformer hot spot temperature is an important indicator of transformer load potential and operational safety, and has been the focus of transformer condition monitoring as well as forecasting. The first disc form the bottom of the winding is in direct contact with the cold oil returned from the radiator, and the surrounding oil temperature is approximately equal to the cold oil, so that its heat dissipation can be analyzed more accurately. In this paper, a new hot spot temperature calculation model based on the principle of transformer heat transfer is proposed. First, based on theoretical derivation, a control equation for the temperature of the first winding disc is set up, and a local thermal circuit model of the first winding disc that can be used for practical operation is established using thermoelectric analogy method. By conducting 11 levels of temperature rise tests with different loads for a total of 21 hours on a 110 kV oil immersed transformer with built-in sensing optical fibers, the temperature changes of the winding under different load rates are obtained. A hot spot temperature model based on the first cake temperature and winding temperature gradient of the winding is proposed. Based on a local thermal circuit model, the first disc temperature of the winding is successfully predicted using load rate and oil temperature at the inlet and outlet of the heat sink as input variables. Subsequently, an exponential and differential equation model of the overall temperature gradient of the winding is established, and the results of the continuous variable load test are predicted based on the proposed hot spot calculation model. The root mean square error of the model prediction is 4.02 K, reducing the maximum error by 21% compared to the traditional model, thereby verifying the effectiveness of the calculation model proposed in this paper.