Improving the computational efficiency of multi-physics simulation and constructing a real-time online simulation method is an important way to realise the virtual–real fusion of entities and data of power equipment with digital twin. In this paper

a data-driven fast calculation method for the temperature field of resin impregnated paper (RIP) bushing used in converter transformer valve-side is proposed

which combines the data dimensionality reduction technology and the surrogate model. After applying the finite element algorithm to obtain the temperature field distribution of RIP bushing under different operation conditions as the input dataset

the proper orthogonal decomposition (POD) algorithm is adopted to reduce the order and obtain the low-dimensional projection of the temperature data. On this basis

the surrogate model is used to construct the mapping relationship between the sensor monitoring data and the low-dimensional projection

so that it can achieve the fast calculation and reconstruction of temperature field distribution. The results show that this method can effectively and quickly calculate the overall temperature field distribution of the RIP bushing. The maximum relative error and the average relative error are less than 4.5% and 0.25%

respectively. The calculation speed is at the millisecond level

meeting the needs of digitalisation of power equipment.