Abstract: The working environment of high-powered power shift tractor is changeable and the working condition is complex, and it shifts gears frequently in the working process. Aiming at the problems of inaccurate pressure control and hysteresis response of wet clutch in tractor power shift transmission (PST), a digital twin driven time-varying proportional integral adaptive control (DT-TVPIA) method is proposed by introducing the concept of digital twin. Firstly, the physical model of wet clutch hydraulic system is established and the interference factors of the physical model are analyzed. Secondly, a digital twin of the wet clutch hydraulic system is constructed using the Input/Output (I/O) data. Thirdly, the Savitzky-Golay filter is used to process the sensor data in real time, and a self-updating controller based on improved projection algorithm and a time-varying proportional integral adaptive controller are designed by considering the control voltage criterion function which minimizes the weighted one-step forward prediction error and introducing the proportional control factor. Finally, the clutch pressure control test based on the rapid control prototype (RCP) is carried out. The results show that pseudo gradient estimates vary with the system state, the digital twin can rapidly update and accurately describe the clutch hydraulic system within 0.445 s when the system state changes dramatically, which proves the validity of the digital twin based on I/O data. Compared with PID, DT-TVPIA has faster clutch pressure response speed and better robustness, and the response time, overshoot and steady-state error of DT-TVPIA target for square wave pressure are within 0.445 s, 0 MPa and 0.001 MPa, respectively, and the response delay time, following error and fluctuation target for sinusoidal pressure are within 0.05 s, 0.0238 MPa and 0.001 MPa, respectively, which meet the pressure control requirements of tractor wet clutch. The research result provide a basis for the improvement of tractor shift quality, and a reference for the application of digital twin in optimal control of complex nonlinear systems.