Controller design for stability and rollover prevention of multi-body ground vehicles with uncertain dynamics and faults.

摘要

Rollover prevention is a fundamental and significant issue for vehicle safety research. Passenger and commercial vehicles with a relatively high center of gravity are especially prone to rollover. Rollover is a threat especially for military vehicles, which operate in severe operational environments and maneuvers. However, many rollover situations cannot be prevented by driver actions alone, even when they are correctly warned. Additional assistance from active anti-rollover control systems can mitigate the deficiency in human capability. Furthermore, rollover events are subject to various perturbations in the rollover parameters (e.g., speed and road adhesion coefficient) and external disturbances (such as adverse weather and terrain conditions). Also vehicles have limited mobility under vehicle component failures resulting from fatigue or field conditions. Hence, the control system has to be fault-tolerant in order to enhance rollover prevention. Thus, with rollover prevention of military multi-body ground vehicles as the main objective of this research, in this dissertation, we first propose a novel control system analysis and design technique by extending the popular Linear Quadratic Regulator (LQR) control design method specializing it for the 'control coupled output regulation' problem. Specifically, in this rollover prevention problem, a 'unified rollover index' is proposed, which captures both the roll dynamics and lateral dynamics, explicitly into the optimization procedure of the LQR framework, which results in a performance index with a coupled term in state and control variables. This LQR design with control coupled output regulation outperforms LQR design with state regulation only, because the cross coupling term helps to prudently allocate the weightings on states and control with the overall performance output minimization as the primary objective rather than individual state regulation. Thus, the proposed rollover prevention technique effectively incorporates the physical nature of the vehicle dynamics into the problem formulation resulting in significantly improved performance. The proposed control design technique is novel and beneficial to the ground vehicle control designers because through this technique, it is shown that the coupling in the vehicle dynamic states and control variables is taken advantage of to improve roll over prevention. In addition, the proposed technique allows us to compare different controller configurations and select the most efficient controller structure in terms of both control effort as well as cost. It is shown that because of the inherent coupling the system has, sometimes it is possible that a well designed single controller (actuator) can result in better performance than multiple controllers (actuators) with improper design. The proposed methodology is illustrated with two applications in the vehicle dynamics area. In the first application, an active steering control system is designed which clearly shows the improved roll over prevention capability of the proposed design compared to the existing designs in the literature. The second application considers a more complicated tractor semi-trailer vehicle and shows how a single active anti-roll bar system at the trailer unit gives better performance than multiple-axle actuators at tractor and trailer together with the double lane change maneuver as the external disturbance. Next the issue of robust control design to handle uncertainties in the vehicle dynamics parameters as well as component faults. Based on the theory of 'Linear interval parameter matrix families', a single robust full state feedback control gain is designed by a convex combination of the control gains designed for finite points (vertices) of the uncertain parameter space. The proposed robust controller design is applied to the multi-body ground vehicle control with uncertainty in the forward speed of the vehicle and the road adhesion coefficient taken into consideration. The results clearly show the efficacy of the proposed robust controller under the assumed perturbations. Thus the proposed techniques in this dissertation help in not only preventing rollover of multi-body ground vehicles with controllers of reduced control effort (which in turn translates to considerable actuator and power savings) but also guarantee the stability and performance for vehicles with uncertain dynamics and faults.