Modelling and identification of car driveline dynamics for anti-jerk controller design

Driveline oscillations, known as ´driveline jerking´, ´driveline judder´ or ´Bonanza effect´ may occur by load changes in modern series-production cars. They cause horizontal accelerations in the chassis and the passenger compartment, which may lessen comfort and driveability to an unacceptable extent. These oscillations are usually reduced by so-called ´anti-jerk functions´ which combine feedback and feedforward control. In practice, the anti-jerk controller parameters are experimentally optimized to combine comfort with sportiness. Such a procedure requires time-consuming driving tests. Alternatively, this paper proposes a model-based approach. The optimal values for feedback control are investigated by simulation in order to reduce time-expensive driving experiments. The simulation is based on a model of driveline dynamics. The model parameters are estimated from measurement data by means of an extended Kalman-filter or a non-linear least-squares optimization procedure. Besides the modelling and identification approach, the results of the model-based controller design are presented. The concept has been validated experimentally on different diesel cars