Advanced model-based control functions to design the longitudinal vehicle dynamics in passenger cars

Abstract

This paper presents a novel approach to design the longitudinal vehicle dynamics with model-based control functions implementable in an ECU. It is shown that an observer, based on a physically motivated model, is capable to capture the longitudinal vehicle behavior reflected by the half shaft torque very well. Thus no additional hardwaresensor is required. Different approaches of observers were evaluated on a real vehicle and obtained good results. Comfort reducing drive train reactions (e.g. drive train oscillations, effect of backlash, etc.) can be minimized by providing an estimated half shaft torque to the control functions. Furthermore this approach allows to implement objective criteria of drivability within online algorithms, to cope with the tradeoff between comfort and agility. Thereupon the acceleration of a passenger car can be directly designed in conformity with human perception within closed-loop control.