To address the problems of difficulty in determination and strong experience dependence of the radius of maximum wind speed and Holland- B parameter in the calculation of the Holland typhoon wind field model, the calculation methods of the radius of maximum wind speed and Holland- B parameter of the Holland typhoon wind field were optimized using the random forest algorithm, combined with the dataset of the machine learning model of satellite remote sensing inversion data. Furthermore, the wind speed of the parametric typhoon wind field model with optimized parameters was compared and verified with the measured wind speed, and the accuracy difference between the parameter values optimized by the algorithm and those calculated by empirical formulas was compared. The results indicate that the two optimized empirical parameters of typhoons fit well with the validation set data; the average errors are both less than 4%, and the prediction accuracy increases by about 50% compared with traditional empirical formulas. Based on the random forest algorithm, the empirical parameters of the parametric typhoon wind field model can be obtained quickly and effectively, and the simulated wind speed error is within 7%. This improves the accuracy and stability of the simulated wind speed of the parametric typhoon wind field model and provides a new technical approach for typhoon numerical simulation.