Vegamoor, Vamsi Krishna (2018-12). Model Based Longitudinal Control of Heavy Duty Vehicles. Master's Thesis. Thesis uri icon

abstract

  • Model based design has proven to be an efficient approach for developing and testing embedded systems. In this work, we attempt to apply this approach to heavy duty vehicles with the goal of implementing a longitudinal controller for velocity tracking. A model was developed in Dymola and parameter identification was performed using specifically designed experiments with a tractor-trailer. The core of longitudinal dynamics of the model involves an engine torque map generated from experimental data. Based on this model, a PID controller was designed and tuned using a closed loop desktop simulation with the plant model. Finally, the controller was implemented in field tests and its performance was verified. We show that this modeling and controller development process can be completed by utilizing the onboard SAE J1939 CAN bus, without the need for any manufacturer privileged information. The model-based controller developed was found to be stable and was able to track a wide range of velocities to within 0.5 m/s (~ 1 MPH) of the desired value. Moreover, the plant model developed in Dymola was confirmed to have sufficient fidelity to be reliably be used for any new control algorithm development in the future.
  • Model based design has proven to be an efficient approach for developing and testing embedded systems. In this work, we attempt to apply this approach to heavy duty vehicles with the goal of implementing a longitudinal controller for velocity tracking. A model was developed in Dymola and parameter identification was performed using specifically designed experiments with a tractor-trailer. The core of longitudinal dynamics of the model involves an engine torque map generated from experimental data. Based on this model, a PID controller was designed and tuned using a closed loop desktop simulation with the plant model. Finally, the controller was implemented in field tests and its performance was verified. We show that this modeling and controller development process can be completed by utilizing the onboard SAE J1939 CAN bus, without the need for any manufacturer privileged information.
    The model-based controller developed was found to be stable and was able to track a wide range of velocities to within 0.5 m/s (~ 1 MPH) of the desired value. Moreover, the plant model developed in Dymola was confirmed to have sufficient fidelity to be reliably be used for any new control algorithm development in the future.

publication date

  • December 2018