Thursday, April 04, 2024

Machine Learning-based realtime calibration of temperature prediction in transmission

Topic and Goal of the Thesis

Similarly to electric machines, gearboxes are also subject to thermal derating. An online state monitoring combined with suitable control strategy can prevent overheating and therefore prolong transmission lifetime. However, that requires highly precise models for temperature prediction incl. correct depiction of thermal inertia to be reliably used by operation strategy optimizer.

One way to increase prediction accuracy is by using machine learning. Therefore, the following thesis aims to develop a plug-andplay calibration algorithm using simulation and measurement data of an e-axle for a commercial vehicle.

Working Points

• Literature research on transmission losses & heat dissipation
• Literature research on measurement concepts
• Literature review on AI & machine learning approaches suitable for real-time applications and / or accuracy improvement
• Familiarization with the software for power loss & temperature calculation
• Implementation and cross-validation of a calibration algorithm

Requirements

• Reliability, commitment, enjoyment of programming and working independently
• Experience with MATLAB and / or Python

Note: Please attach brief resume and grade summary.