It is necessary to obtain the lateral and torsional stiffness of aviation tire to analyze the shimmy of landing gear. In this pa?per， a set of intelligent prediction methods of aircraft tire lateral and torsional stiffness are established and successfully applied in the anti-shimmy design of landing gear of a UAV. 503 groups of tire lateral and torsional stiffness test data were collected， including 3 tire manufacturers， 7 tire models， different combinations of air pressure， load， compression， diameter and width. A fully connected deep learning neural network model is established. The neural network model is trained through the existing tire stiffness test data. From the training effect of the randomly assigned verification set， it can be seen that after 1000 times of training， the mean error basically converges， and the predicted value is basically consistent with the real value. The trained model is used to predict the lateral and torsional stiffness of the tire of a UAV， and the prediction results are applied to the analysis of the shimmy of the landing gear to determine the critical damping of the landing gear and guide the design of the shimmy damper.