Abstract
Keywords - Slow-Active suspension, LQR algorithm, Preview, Digital control
Abstract - The study in this paper presents the development of a band-limited controller for the Slow-Active suspension system utilizing preview information by using co-simulation technology, which provides the rapid iteration of the control system and insight into its effects on vehicle dynamics even at the early stage of the design process.
Firstly, the vehicle dynamic model is built by using the ADAMS/View software package, which is mainly considering the vertical and pitch motion of the body, and the vertical motions of tires. Meanwhile, the linear model based on the equations of motion which describes the behavior of the vehicle model is also established. The validity of the multi-body model and the linear model can be verified by each other for their identical frequency responses. Then, a discrete time optimal control algorithm is proposed for the design of the limited bandwidth active suspension system. In order to maximize the performance potential, both look-ahead and wheelbase preview information are utilized. The controller design is based on the Linear Quadratic Regulator (LQR) algorithm. Once the controller design is completed and the validity of the model is guaranteed, the control loop can be properly integrated with the multi-body dynamic vehicle model seamlessly by the Control Interface and therefore the simulation will be initiated from the familiar Simulink environment. Based on the data communication between the vehicle dynamic model in ADAMS and the designed controller in Simulink during simulation, the control algorithm can be iterated until the satisfactory simulation results are available and the benefit brought by preview information can be clearly investigated.
Current study showed that the above approach can be utilized effectively in the analysis and design of active suspension and the subsequent results are reliable and promising, which will enable us in the future to refine our model and extend our current work more closely to the reality.