Q-learning based adaptive PID controller design for AMT clutch engagement during start-up process

Start-up is a very important process for vehicle, and the handling process of the start-up or "launch" by the driver from the beginning to handle to master proficient skill is a learning process. This paper presents an adaptive PID controller for start-up process of AMT vehicle by Q-learning algorithm. The switch strategy of PID parameters is obtained by considering the contradictive requirements of less friction loss and less driveline shock during clutch engagement. The proposed control strategy is tested on a powertrain simulation model. Through a large number of simulations, it is verified that the system with the designed controller has ability to deal with the uncertainties, such as vehicle mass and road grade.