A Flexible Hierarchical Model-Based Control Methodology for Vehicle Active Safety Systems.

Abstract

To improve the stability and safety performance of active safety systems, vehicle control systems are increasingly incorporating sophisticated chassis actuator control functions. Thus vehicle control systems require efficient control algorithms to reduce the amount of duplicate hardware or adopt various actuator combinations in response to diverse customer demands and various hardware combinations from different suppliers. One of the main challenges for efficient controller design is how to create a flexible modular design which can respond to different reference vehicle behaviors and provide optimal actuator apportionment, while avoiding functional conflicts between chassis sub-systems. The proposed control methodology offers a total vehicle motion control solution for an active safety system by addressing the issue of flexible modular design in integrated vehicle control systems. In this thesis, modularity is realized by a model-based hierarchical control structure consisting of three layers: an upper layer for reference vehicle motions, an intermediate layer for actuator apportionment, and a lower layer for stand-alone actuator control. The reference vehicle motions can be determined by any type of reference model for vehicle stability control or collision avoidance control. The actuator apportionment uses Model Predictive Control (MPC) to provide flexibility by balancing tire forces to track target reference vehicle motions while simultaneously considering constrained conditions, such as actuator limits or available actuator combinations. Moreover, the MPC is designed to be feasible in real-time using a linear time-varying MPC approach which avoids the complexity of full nonlinear MPC and addresses the vehicle nonlinearity. The effectiveness of the proposed control structure is investigated for vehicle stability control in terms of handling stability and handling responsiveness. The handing stability and responsiveness of the control structure appear to be robust with respect to various uncertain environments including model-plant mismatch and diverse driving conditions. It is then applied to collision avoidance systems by adapting the reference vehicle motions at the upper level of the controller. Collision avoidance is found to be nearly as effective as that of the combination of an optimizing driver supported by the MPC-based stability controller; this suggests that the MPC approach could be used in future high performance collision avoidance systems.