Semi-Active Nonlinear Predictive Suspension Control for Off-Road Vehicles

Abstract

Off-road vehicle suspension systems are needed for the safe and comfortable traversal of off-road terrain. The use of semi-active suspension for off-road has previously been only considered using existing on-road techniques. However, the behaviour of tyres on soft-soils changes considerably when considering the sinkage of the tyre into the soil. The focus of this thesis is the proposal of a nonlinear model predictive controller which is capable of being used for the control of semi-active magnetorheological damper based suspension on off-road soft
soils. This thesis proposes modifications to existing nonlinear damper models to improve the accuracy of the damper forces used during simulations. Utilising the newly modified damper
models this thesis proposes formulations of two vehicle models; quarter car, and full car. These formulations are utilised in the composition and simulation of a simulated nonlinear model
predictive controller for on-road performance assessment against other controller formulations. A computationally effective controller is also proposed, namely an adaptive proportional-integral-derivative controller. It adjusts its gains in proportion to the normalised suspension deflection and relative velocity. It is combined with cascade control techniques for the case of
full car suspension controller. The controllers are tuned to manipulate the current supply to a magnetorheological damper to achieve a reduction in the vertical accelerations of the vehicle
body in order to improve vehicle ride comfort. The nonlinear model predictive controller was also modified and utilised for off-road soft soil suspension control on both flat and gently sloping terrain. The soft soil behaviour of the tyre is modelled using a modified tyre-soil interaction model which is adapted here to improve the speed of calculation through the use of lookup tables. The simulation studies conducted herein show that the nonlinear model predictive controller can perform far better than the proportional-integral-derivative (PID) controllers when operated on-road due to its ability to predict the behaviour of the vehicle response to incoming disturbances. The PID based controllers are tuned for a specific scenario. The nonlinear model predictive controller have the advantage of being able to adapting to a change in scenarios. However, the improvement seen for the controller on soft soils is marginally better and suggestions for further research into both the tuning process for the controller and the method of integration between the tyre-soil interaction model and the vehicle model are proposed in the conclusion.

Date of Award	2022
Original language	English
Awarding Institution	Coventry University
Supervisor	Olivier Haas (Supervisor) & Mike Blundell (Supervisor)