A dynamic model of the starting process for a wet dual clutch transmission (DCT) vehicle, reflecting changes in clutch friction coefficient and the attenuation of clutch performance, is established. Taking minimum jerk intensity, friction work and starting sliding time as optimization objectives, the optimal transmission torque of the clutch during the vehicle starting process is determined by using the linear quadratic regulator (LQR). To address the impact of clutch performance attenuation and the variation of clutch friction coefficient with the sliding velocity on clutch pressure control, a nonlinear robust control strategy for clutch pressure is proposed. This strategy is designed to track the optimal transmission torque obtained by LQR. Results show that the proposed strategy can effectively track the optimal transmission torque even under changes in clutch friction coefficient, with an average tracking error of about 0.02 N·m. In comparison with the PID control strategy, it exhibits a more accurate control effect and greater robustness.