Abstract
Keywords real-time simulation, vehicle dynamics, neuro-fuzzy model, active steering system, controller design
Abstract A new approach for an active steering control system based on local linear models to improve the lateral dynamic behavior of passenger cars is considered. In combination with braking control systems the active steering system is used to stabilize the vehicle in critical situations.
Similar to the Electronic Stability Program (ESP) which uses active braking, the active steering system compares the desired yaw rate calculated from the steering wheel angle with the actual yaw rate. To calculate the desired yaw rate a time-variant one track model is used for a dry road surface. A feedback controller is designed to control the desired yaw rate given by the driver by generating an additional steering wheel angle. The used one track model depends on the vehicle velocity, on front and rear cornering stiffness, on the vehicle mass and on the moment of inertia. Furthermore, dynamic vertical load variation, large tire side slip angles and the influence of roll and pitch motions result in deviations of the one track model. However, the vehicle velocity can be measured and its influence on the parameters of the one track model can be considered in a local linear model approach. The velocity range is divided into sections in which the behavior of the vehicle is approximately linear. For each section a controller can be designed with conventional linear design tools. The overall controller is obtained by the superposition of the local linear controllers with Gaussian weighting functions.
With this approach the time-variant influence of the vehicle velocity can be approximately eliminated and the resulting model used for controller design is more precise over a large velocity range than with one single linear model and constant parameters. Therefore the controller performance can be improved, especially for driving maneuvers with fast changes in the vehicle velocity (e.g. braking).
The controller is tested for a critical driving situation using a Hardware-in-the-Loop (HIL) simulator. Disturbances are introduced by braking on µ-split road surfaces. The HIL simulator is composed of hardware (steering wheel) and software (vehicle simulation model). This arrangement allows to test the controller close to reality. Furthermore, the interaction of driver and controller can be taken into account when evaluating the developed control system. The benefits of the presented local linear controllers are demonstrated in form of simulation results and 3D animations.