1. BaSaMa & HiWi
    2.  

LLM-Based Error Detection in the Implementation of Battery Specifications in Testing Processes for E-Mobility

Institute
Lehrstuhl für Fahrzeugtechnik (TUM-ED)
Type
Semester Thesis / Master's Thesis /
Content
theoretical /  
Description

Motivation:

The electrification of the mobility sector places new demands on vehicle development and validation. Lithium-ion traction batteries (LIBs) are the most expensive and most critical component in this context. Their comprehensive testing—particularly with regard to aging behavior—is essential.

Customer requirements are defined in the form of test specifications and transferred into test plans. In this process, formal and semantic errors frequently arise, which are difficult or even impossible to detect without significant effort, and which severely impair the quality of the test results.

Topic of the Thesis:

Within the TwinBat project, a collaboration with the project partner TÜV Süd Battery Testing GmbH aims to develop an LLM-based method that automatically detects errors in the implementation of specifications—either in a test plan before testing begins or in the measurement data after testing has been carried out.

Scope / Content of the Thesis:

  • Parsing and semantic interpretation of test specifications

  • Alignment of specification and test program

  • Evaluation of measurement data against the specification

  • Clear communication of detected errors to test engineers

The objective is an AI-supported, end-to-end validation chain for battery testing processes.

Your Benefits:

  • Contribution to sustainable mobility

  • High practical relevance and industrial value

  • Strong reference for career entry

  • Work at the intersection of AI, e-mobility, and engineering

  • In case of excellent performance: co-authorship on a scientific paper

Work Packages:

  • Literature review on LLMs in the context of the target application and the processing of sensitive data

  • Selection of suitable LLMs

  • Development of an error-detection framework

  • Validation using test cases from the TwinBat project

  • Documentation and critical discussion of the results

Requirements

Requirements:

  • Interest in e-mobility and lithium-ion batteries

  • Ability to work independently

  • Very good German or English language skills

  • Fundamentals in AI / machine learning

  • Programming skills in Python

Tags
FTM Studienarbeit, FTM EV, FTM EV Powertrain, FTM Brehler
Possible start
sofort
Contact
Tobias Brehler, M.Sc.
Phone: +49 89 289 15782
tobias.brehlertum.de