Model Predictive Control for Real-Time Simulation of a Network-Controlled Vehicle Drivetrain

This paper proposes a real-time implementation of a networked predictive controller designed to damp driveline oscillations, which is crucial in improving drive ability and passenger comfort, while compensating the time-varying delays that appear due to sending the control commands and the measurements from the sensors through Controller Area Network (CAN). Firstly, the designed real-time structure integrated with CAN test-bench is described and then the model of the drive train is derived. Secondly, considering that the CAN-induced time-varying delays are bounded, a method to model the physical plant (vehicle drive train) including the delays is proposed. Then, a predictive control strategy, which makes use of the previously developed model, is designed in order to damp the driveline oscillations. The proposed control scheme is tested using the designed test-bench and the experiments based on realistic scenarios show that the proposed controller can outperform classical controllers, e.g., PI.