Robustness Analysis of Look-ahead Control for Heavy Trucks
This thesis work provides an approach to analyse the robustness
of a fuel-time-optimal controller for the longitudinal dynamics
of heavy trucks. The analysed look-ahead control system uses a
predictive control strategy, developed in a collaboration of
Scania and Linköpings Universitet. It utilizes GPS positioning
data and a road slope database to compute the optimal fueling,
braking and gear choice. In this study, the robustness towards
parametric uncertainties is tested in various simulations on a
Matlab/Simulink model. The experiment vehicle, which is modeled,
is a Scania tractor with semitrailer of 15 to 40 tonnes with a
310Hp engine. The main focus within these tests is on
uncertainties of the vehicle mass.
The simulation model is based on the evaluation model from
[Hellström et. al 2010], which was used to simulate the
look-ahead control prior to and after practical experiments. The
model has been modified to include perturbations of the vehicle
mass and positioning data. The various simulations consist of a
set of runs on a modeled 120km long motorway link between the two
Swedish cities of Norrköping and Södertälje and the uncertainties
are being analysed by changing one factor at a time in the
simulation model.
In conclusions, the analysed controller acts very robust towards
uncertainties in mass, though the effects of a wrong estimated
mass get bigger at lower masses. At 15t of true mass, an
estimated mass over 19t would lead to a fuel-time consumption
that is higher than the comparable cruise controller result. A
true mass of 20t does already require an estimated mass over 40t
to get insufficient results. Lower estimated masses result in the
worst case in the same fuel-time use as the cruise controller. At
higher true masses all tested estimated masses cause satisfactory
fuel consumption and trip time. However, since the estimated mass
is not very likely to exceed ±10% of the true mass, the
controller should be practically robust in all tested scenarios.
Uncertainties of the GPS position have been analysed in a large
range and the results display the controller very robust towards
these errors as well. The algorithm is even with bigger
uncertianties capable of lowering the fuel consumtion without
increasing the trip time.
Wolf Krahwinkel
2010
Informationsansvarig: webmaster
Senast uppdaterad: 2021-11-10
LiU Homepage
