The invention relates to the field of automatic transmissions of automobiles, and specifically relates to an intelligent predication method based on a support vector machine algorithm and used for theclutch torques of a dual-clutch transmission. The intelligent predication method based on the support vector machine algorithm and used for the clutch torques of the dual-clutch transmission comprises the following steps: 1) obtaining the torque data of two clutches in a gearshift process by virtue of a fuzzy control method for the clutch torques; 2) carrying out multiple Fourier fitting on a plurality of data points obtained through simulation, and obtaining the formulas of the torque curves of the two clutches separately; 3) optimizing the clutch torques by virtue of a genetic optimizationalgorithm to obtain the optimized torque curves, and providing label data for training a model based on the support vector machine algorithm; 4) establishing the intelligent predication model based onthe support vector machine algorithm and used for the clutch torques, and training; and 5) compiling the trained predication model based on the support vector machine algorithm and used for the clutch torques into an automatic transmission control unit loaded in a real vehicle, and obtaining the optimal target torques of the clutches to control a gearshift execution mechanism according to the rotational speeds of drive ends and driven ends of the clutches, which are measured in real time.