The tasks of the thesis include the analysis of safety-critical thermal runaway scenarios in lithium-ion batteries and the development of a methodology for AI-supported detection, tracking, and evaluation of particles in the venting process. For this purpose, suitable neural network architectures are identified for particle tracking in highly dynamic and optically challenging environments, and their performance is evaluated in comparison to established image-based methods such as PIV. Based on experimental high-speed recordings, relevant physical parameters are derived and their validity for characterising the thermal runaway event is assessed. The work typically includes the following focal points:

Literature review on particle tracking procedures, image-based flow analysis, and AI methods in the context of safety-critical battery scenarios (e.g., neural networks, deep learning for computer vision, tracking algorithms). Analysis of the physical and measurement-related boundary conditions at thermal runaway of lithium-ion batteries, particularly the gas particle emission and the associated thermal, mechanical, and abrasive stresses. Design, implementation, and training of a neural network for detection and tracking of particles in high-speed recordings of the venting process. Evaluation and validation of the developed approach using experimental data and comparison with classical PIV methods regarding accuracy, robustness, and validity. Derivation of characteristic parameters for qualitative and quantitative description of particle emission during thermal runaway. Documentation and scientific preparation of the results in the framework of a bachelor's or master's thesis with strong practical relevance.

XML job scraping automation by YubHub