Dynamic Modelling and Predictive Control of Nonlinear Electronic Throttle Systems

The transition from mechanical to electronic throttle control has improved the efficiency and adaptability of stoichiometric combustion engines. However, nonlinear dynamics are introduced by friction, return spring, and parameter variations, making it challenging to achieve a fast and precise position control.This thesis focuses on dynamic modelling and implementation of control strategies for an electronic throttle system, with the aim to evaluate predictability of position control. A dynamic model of the throttle system is developed, a digital twin, with nonlinearities such as Coulomb friction and the limp-home feature. System parameters are identified using test bench experiments to estimate static friction. Dynamic parameters in the model were estimated using optimisation techniques. With these parameters a digital twin is made, ensuring an accurate representation of the test bench with an actual throttle. A PID controller with nonlinear compensators and three design variants of an active disturbance rejection control with and without nonlinear compensators are implemented and evaluated. The performance of the controllers is assessed through hardware testing and simulation. The results demonstrate that the digital twin model performed similar to the actual model but a non modelled stick-slip friction was identified during test in HIL. The results from the control models showed that the ADRC controller with a feed forward term compensating for the nonlinearities in the system had the highest predictability.