Modeling of a Motorcycle in Dymola/Modelica
Abstract
This thesis introduces a new and freely available Modelica library for the purpose of
simulation, analysis and control of bicycles and motorcycles (single-track vehicles). The
library is called
MotorcycleLib and focuses on the modeling of virtual riders based on
automatic controller design. For the vehicles, several models of different complexity
have been developed. To validate these models and their driving performance, virtual
riders are provided. The main task of a virtual rider is to track either a roll angle
profile or a pre-defined trajectory using path preview information. Both methods are
implemented and several test tracks are also included in the library. Regarding the
stability of uncontrolled vehicles, an eigenvalue analysis is performed in order to obtain
the self-stabilizing area. A key task for a virtual rider is to stabilize the motorcycle.
To this end, the controller has to generate a suitable steering torque based on
the feedback of appropriate state variables (e.g. lean angle and lean rate). One major
problem in controlling two-wheeled (single-track) vehicles is that the coefficients of the
controller are strongly velocity dependent. This makes the manual configuration of a
controller laborious and error-prone. To overcome this problem, an automatic calculation
of the controller's coefficients is desired. This calculation requires a preceding
eigenvalue analysis of the corresponding uncontrolled vehicle. This enables a convenient
controller design, and hence several control laws that ensure a stable driving behavior
are provided. The corresponding output represents a state feedback matrix that can
be directly applied to ready-made controllers which are the core of virtual riders. The
functionality of this method is illustrated by several examples in the library.
Interested in reading the
entire thesis?
(208 pages, 5,857,877 bytes, pdf)
Deutsche Zusammenfassung
Homepage
Last modified: September 23, 2009 -- © François Cellier