Abstract: With the rapid growth of wind power installed capacity, the reliability of insulation performance of box-type transformer windings directly affects the operation safety of wind turbines. Aiming at the insulation performance evaluation of box-type transformer windings of Yuanjing wind turbine, this paper proposes an insulation performance diagnosis method based on multi-parameter fusion. By establishing a comprehensive evaluation model including dielectric loss, partial discharge and temperature characteristics, accurate evaluation of winding insulation status is achieved. Experimental results show that the dielectric loss measurement accuracy of the proposed method reaches ±0.00005, the partial discharge positioning error is less than 5mm, and the aging state recognition rate is 98.2%. Compared with the existing deep learning-based methods and multi-physical field coupling analysis method, it has obvious advantages. This research provides a reliable technical means for condition-based maintenance of box-type transformers in wind farms, and is of great significance to improving the operating reliability of wind power.