Electronic throttle state estimation and hybrid theory based optimal control

Electronic throttle body (ETB) is a car device that regulates air inflow into the motor´s combustion system. Its performance has a major impact on the quality of the overall engine speed control. However, due to the usage of cheap components and some design features the ETB exhibits several nonlinear phenomena. This nonlinear behavior and the fact that there is only one measurement available - low quality measurement of the throttle position makes a design of the throttle controller a challenging task. Our approach is to model the ETB as a discrete time piecewise affine (PWA) system and apply model predictive control (MPC) strategy to design an explicit state feedback control law. Since MPC is a full-state controller and there is only one measurement available the rest of the states have to be estimated. We have chosen unscented Kalman filter (UKF) for the estimation purpose since it was performing the best in the presence of strong, almost discontinuous, process nonlinearities. In the end, MPC and UKF algorithm were implemented and tested on the real electronic throttle for the case of set-point reference. Experimental results indicate that the performance of cheaply produced components can be significantly improved with a good control strategy.