State estimation of automotive drive with control applicationsDoctoral thesis (in Croatian), Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb, Croatia, 2007. , 2007
The thesis deals with the estimation of state variables of different automotive power train subsystems, and its control applications. For that purpose appropriate dynamic models of different power train subsystems have been considered, such as the spark-ignition engine model and tire-road friction model. The model parameters have been determined by means of experimental identification. The emphasis has been given to experimental characterization of tire friction behavior in the low wheel slip region. This includes the estimation of tire static curve gradient and tire vibration mode damping ratio for different road conditions, and correlating these parameters with tire friction coefficient for a wide range of vehicle operation. An adaptive Kalman filter, which is based on power train subsystem stochastic state-space model, is proposed for the purpose of estimation of automotive power train variables. The basic Kalman filter is tuned for good noise suppression, while the short-duration adaptation of Kalman filter gains is performed only when sudden changes of state variables are detected. The special emphasis is given to adaptive Kalman filter design for SI engine load torque estimation. The adaptive load torque estimator is characterized by good tracking ability of fast load torque changes and low noise sensitivity. The proposed estimator has been used as a basis for SI engine load torque compensation within the engine idle speed control system. Such an adaptive controller has been verified experimentally, and compared to conventional controllers such as PI, PID, and polynomial controllers. The comparative experimental results point out to far better control performance of adaptive controller for a wide range of SI engine operation. The experimental analysis also shows that the adaptive compensator application yields a kind of ultimate performance for the considered case when the load torque cannot be accurately measured or reconstructed. The adaptive Kalman filter has also been successfully applied for the estimation of SI engine air mass flow and electrical vehicle tire traction force. The application of adaptive Kalman filter for the estimation of tire-road friction coefficient based on the tire static curve gradient and the vibration mode damping ratio results in accurate and fast detection of road condition change, and good noise suppression.