Supervisory Control of a Dual-Direct-Drive Electrical Vehicle

Studenteropgave: Kandidatspeciale og HD afgangsprojekt

  • Bjarke Kleif
  • Steffen Thestrup Rasmussen
As part of an electric vehicle research project this MSc thesis investigated fully- and
semi-autonomous control strategies to be implemented on a modified 1:5 scaled RC
car with electric motors powering each rear wheel.
The powertrains were powered by BLDC motors which were controlled by
prefabricated motor controllers. System identification revealed a linear input/output
relationship and the motors were therefore modelled as DC motors. It was desired
to control the output torques and therefore PI control with torque feedback was
implemented on each powertrain.
A vehicle simulation platform was developed in Simulink R
using the full car
lumped mass model approach. Wheel traction forces were modelled using general
formula found in automotive literature, in particular the magic formula was used for
longitudinal traction forces.
An automatic test procedure controller was derived which consisted of a yaw-rate and
speed controller. The yaw-rate controller was developed using feedback linearization
with linear PI control and the speed controller was developed using wheel speed
feedback with PI control. A semi-autonomous DFA supervisor switched between
control algorithms according to the driving situation.
Different strategies to dispatch torque were investigated. One dispatcher which
assisted the driver in taking the desired turn showed promising results, and could
prove more effective than an open differential.
State estimation of yaw-rate, vehicle velocity, road incline and longitudinal wheel slip
was performed and implemented in the simulation platform and on the RC platform.
Furthermore a novel approach for torque estimation on BLDC motors was presented
using two phase currents and HES, which showed promising results compared to other
more hardware requiring methods.
Stability of all controllers was shown under the assumption of vanishing perturbations.
Individual- and integration tests in the simulation platform revealed good performance
of all controllers derived. Moreover, traction control was implemented on the RC
platform and the proposed method was verified.
Udgivelsesdato15 jun. 2012
Antal sider249
ID: 63440971