Abstract: In order to ensure the safe and stable operation of transformers and improve the accuracy of transformer winding hot spot temperature prediction,a transformer winding hot spot temperature prediction method based on principal component analysis-improved Harris hawks optimization-least squares support vector machine（PCA-IHHO-LSSVM）was proposed.Principal component analysis was conducted on the main characteristic variables for predicting the hot spot temperature of transformer windings.The cumulative contribution rate of the three characteristic variables,active power,load current,and top oil temperature,exceeded 85%. The three principal components of active power,load current,and top oil temperature were determined,and the input variables were reconstructed as indicators. Using Tent chaotic mapping and student distribution perturbation strategy,the Harris Eagle algorithm was improved to improve the convergence accuracy and optimization performance of the IHHO algorithm. The IHHO algorithm was used to optimize the parameters of LSSVM,and a transformer winding hot spot temperature prediction model based on PCA-IHHO-LSSVM was established. The simulation analysis was carried out with the transformer monitoring data,and compared with other winding hot spot temperature prediction methods.The results show that the average relative error and Root-mean-square deviation of the PCA-IHHO-LSSVM model proposed are2.92% and 1.77 ℃ respectively. The prediction accuracy is higher than other methods,which verifies the practicability and superiority of the transformer winding hot spot temperature prediction proposed in this paper.