Abstract
ABSTRACT:
The last few years have seen scientific and technical competition within the automobile
industry become increasingly intense. Because of this it is important for individual manufacturers to ensure that new designs are brought to the market as quickly as possible. In the future it must be possible to carry out the design of a new model, from conception to production, in as little as eighteen months.
These days the equipment within a motor vehicle includes a myriad of electrical and electronic subsystems. Many of these subsystems serve to improve driving comfort, i.e. electric windows, air conditioning system, etc. On top of this there are safety relevant vehicle subsystems, such as drive dynamics control and anti-lock brake
systems.
An ever decreasing design effort for such systems, coupled with increasing amounts of equipment, is only possible due to the far reaching use of computer simulations in the design of new vehicles. The aim of computer models is to reveal, as early as possible in the design phase, the effect on the dynamic behaviour of the vehicle of new components operating in conjunction with the existing subsystems. With such an approach the effect of a new component can be analysed in the definition phase, long
before the prototype is complete.
To date, modelling efforts have concentrated on reproducing as exactly as possible the behaviour of individual components. This approach has yielded, for example, exact descriptions of the wheel dynamics using the finite element method. Simulations of such models are computationally expensive and time consuming. Due to the fact that
there is a huge difference between different vehicles in terms of structure and kinematics, such models are very specific, and special know- how is required to alter a model for use with a different vehicle type.
The modelling approach has the following aims: reduction of model complexity to a level sufficient for vehicle dynamics; implementation on a PC in a common programming language, so that usage is widely spread instead of being constrained to a few specialised departments; interaction between submodels, whereby the design
time is concentrated upon the subsystem currently under investigation; only necessary accuracy, so that time consuming tests with experimental vehicles can be reduced.
In order to obtain these aims, the system will be decomposed into its individual components. When carrying out such a partitioning, it is important to ensure that meaningful variables are chosen for the interfaces between the different submodels. It is thus sensible to choose, insofar as it is possible, torques or forces and angular or longitudinal velocities.
In general, observers are implemented when certain state variables cannot be measured without unacceptable expense. If, as is here the case, the system is in nonlinear form, the direct application of a Luenberger observer is not possible. To overcome this problem in the paper a nonlinear observer is designed and used for the
observation of the vehicle body side slip angle, based on reduced nonlinear two-track model.