Integrated torque-vectoring and anti-roll moment distribution strategies based on optimal control: influence of model complexity and road curvature preview

N. H. AmerM. DalboniP. GeorgievC. CaponioD. TaverniniP. GruberM. DhaensA. Sorniottia Centre for Automotive Engineering,University of Surrey,Guildford,UKb Centre for Defence Research and Technology (CODRAT),National Defence University of Malaysia,Kuala Lumpur,Malaysiac Department of Engineering and Architecture,University of Parma,Parma,Italyd Research Department,Tenneco Automotive,Sint-Truiden,Belgiume Department of Mechanical and Aerospace Engineering (DIMEAS),Politecnico di Torino,Turin,Italy
DOI: https://doi.org/10.1080/00423114.2023.2296586
IF: 3.749
2024-01-16
Vehicle System Dynamics
Abstract:Although the vehicle dynamics effects of variable anti-roll moment distribution actuated through active suspension systems are widely discussed in the literature, their model-based control has only been recently analysed, given the highly nonlinear nature of the involved dynamics. Moreover, the available studies do not discuss the trade-off between internal model complexity and controller performance, nor analyse the opportunities offered by vehicle connectivity, which enables the prediction of the steering angle and reference yaw rate profiles ahead. To address the gap, this paper introduces and assesses three optimal controllers for an electric vehicle with active suspensions, multiple powertrains, and a brake-by-wire system. The formulations are: (a) a gain scheduled output feedback linear quadratic regulator (OFLQR); (b) a nonlinear model predictive controller using a three-degree-of-freedom prediction model, without and with preview of the steering angle and reference yaw rate ahead, respectively referred to as NMPC-3 and NMPC-3-Pre; and (c) a nonlinear model predictive controller based on an eight-degree-of-freedom prediction model, referred to as NMPC-8 and NMPC-8-Pre depending on the absence or presence of preview. The results on an experimentally validated model show that: (i) NMPC-8 provides evident yaw rate tracking benefits with respect to (w.r.t) OFLQR and NMPC-3; and (ii) NMPC-8-Pre can bring ∼20% yaw rate tracking improvement w.r.t. an optimally tuned NMPC-8 configuration.
engineering, mechanical
What problem does this paper attempt to address?
The paper attempts to address the trade-off between model complexity and controller performance in vehicle dynamics control, as well as the impact of vehicle connectivity (such as road curvature preview) on control strategies. Specifically, the study aims to: 1. **Analyze the relationship between model complexity and controller performance**: Explore the performance of predictive models with different complexities (e.g., 3 degrees of freedom and 8 degrees of freedom models) in vehicle dynamics control. 2. **Evaluate the impact of road curvature preview on controller performance**: Investigate how road curvature information obtained through vehicle connectivity can improve control effectiveness, especially in the context of previewing future driving paths. 3. **Design and evaluate an integrated direct yaw moment and anti-roll moment distribution controller**: Propose and evaluate three optimal control-based controllers, namely: - Gain-scheduled output feedback linear quadratic regulator (OFLQR) - Nonlinear model predictive controller based on a 3 degrees of freedom predictive model (NMPC-3), without and with road curvature preview (NMPC-3-Pre) - Nonlinear model predictive controller based on an 8 degrees of freedom predictive model (NMPC-8), without and with road curvature preview (NMPC-8-Pre) Through experimentally validated models, the research results indicate that: - NMPC-8 outperforms OFLQR and NMPC-3 in yaw rate tracking. - NMPC-8-Pre with road curvature preview improves yaw rate tracking performance by approximately 20% compared to the optimally tuned NMPC-8 configuration. These studies contribute to enhancing the handling stability and safety of electric vehicles, particularly under complex driving conditions.