Abstract: Aiming at optimizing the warpage deformation of PA6/GF30 intake manifold to improve the air intake performance of the manifold， a response surface experiment was designed by using Box-Benhnken， and then a random forest regression model was constructed. The data obtained from the response surface experiment were used as the original data for optimal prediction. The results indicated that the predicted values of the warpage deformation of the manifold exhibited a good fit with the actual values， suggesting that the random forest regression model was effective. The air intake process of the manifold was simulated by means of the Moldflow-Ansys joint simulation. The simulation results showed that the warpage deformation of the optimized manifold decreased from 1.413 mm to 1.032 mm， which was reduced by 27 %. The deviation rate and unevenness of the intake air flow decreased from 7.498 3 % to 4.242 8 % with ［-3.944?153?2 %， 3.554?111 %］ and ［-2.545?082?2 %， 1.697?694?7 %］， respectively， indicating a significant improvement in air intake performance.

Key words: intake manifold, Box-Benhnken, response surface, random forest, co-simulation