Accurate suspension models for active vehicle control synthesis using model reduction techniques.

摘要

Low-order linear models have been used mostly for advanced vehicle control systems because of their simplicity. The two-mass model and the bicycle model are typical examples for designing active suspensions or four wheel steering systems. In general, the equivalent parameters of these models are assumed to be available and provided by the component data. However, such nominal models may not be accurate enough without considering the influence of the suspension's kinematic structure.; An accurate, yet simple model is desired for minimum tuning of control gains as well as reduced computation and measurement.; Two approaches have been applied to obtain such a model: (1) using a parameter identification method with the given simple model structure, (2) using model reduction techniques to simplify a large-order MBD (multi-body dynamics) model.; For a quarter car system, the difference between the identified parameters and the nominal parameters shows the existence of parameter uncertainty due solely to the suspension linkage layout. The model reduction techniques resulted in a more accurate set of equivalent parameters. The results also show the effectiveness of the two-mass model when the equivalent parameters are properly provided.; A large-order linear model is obtained by linearizing a MBD model for a full car suspension system. The states of the model are divided into two groups depending on their effects on the ride and handling performances. The singular perturbation method is; then applied to reduce the model size. Two accurate reduced-order models for the ride and handling analysis for a full car system are obtained.; The linear active systems based on these accurate simple models exhibit improved performance over the nominal models. Additionally, the accurate models aid control designers in the selection of optimized control gains. In the case of sliding mode controls, the accurate models also provide efficient controllers with reduced switching gains.; In this way, the study offers a critical link facilitating the concurrent engineering of suspension design and control synthesis. It also provides an analytical method to investigate the effects of model complexity on model accuracy for vehicle suspension systems.