Driveline SciML Demo ​
Download this exampleDrivelineSciML.zipFull project — Dyad model + Julia + data
An electric car has a rubbery coupling between its motor and its wheels that soaks up shock. Nobody has measured how it behaves, so this demo shakes it in controlled ways and fits a model to what comes back.
That coupling is a nonlinear torsional isolator: a twisting spring and damper whose stiffness depends on how fast and how far it is twisted. Its torque comes from three paths at once:
a baseline path — how it responds to slow, steady twisting;
a Maxwell relaxation element — extra stiffness that shows up only at high frequencies;
a Bouc-Wen hysteresis element — torque that depends on where the twist has been, not just where it is now.
The demo recovers the hidden parameters in two stages. Stage 1 shakes the coupling gently across a sweep of frequencies; Stage 2 cycles it slowly at full amplitude. Both stages fit with SciML optimization.
Note
This is a heavy SciML demo. The model diagram below renders from a snapshot; run the calibration itself from the DrivelineSciML project.
The model ​
DrivelineSystem is the assembled driveline: the torque source, engine inertia, Maxwell-Bouc-Wen isolator, load inertia, load damper, and ground: