Extracting Knowledge on Seal Wear through Unsupervised Anomaly Detection

  • Background

    In today's research and development processes, life cycle testing of critical components is crucial to ensure reliable and sound products and services. One example of such components are sealings used in the automotive or aerospace industry. Since the effort and cost factor of humans checking the usage characteristics (e.g., wear) of seals is high, artificial intelligence (AI) as an emerging technology can help to accelerate testing and development as well as improve product quality by revealing new knowledge on preceding processing steps. To do so, sensors are installed on machines and used to collect data relevant to analyze the seal quality.


    Research Goal

    In cooperation with Trelleborg Sealing Solutions, you will apply state of the art machine learning techniques to develop an unsupervised anomaly detection artefact that can assist domain experts in analyzing seals and generate new knowledge on the part’s quality. With the ability of ML models to uncover new patterns in the data the causal analysis on seals’ quality can be improved. Through this collaboration of domain expert and ML model, not only does the robustness of the entire process advance. In addition, domain experts benefit from new insights and can increase their own know-how on seal anomalies’ root cause. In this, explainable artificial intelligence can facilitate to make the ML model more transparent and support to reveal its own knowledge on the seals’ quality. Moreover, you will implement a prototype in Python to evaluate the artefact on real data and compare the quality of worn seals.


    Working on this thesis you will:

    - Apply your theoretical knowledge to a practical use case in the context of sealing systems
    - Work with sensors data from research and development
    - Work directly with domain experts and apply state-of the art machine learning techniques


    We look forward to receiving your application because you:

    - You are interested in the field of machine learning
    - You are highly motivated to work on recent real-world problems in a self-organized and goal-oriented working mode and you bring in own ideas
    - You are open-minded and don’t hesitate to get in touch with professionals
    - Very good English skills as the thesis will be written in English



     - Start: Immediately

     - Duration: 6 months

     - Location: Trelleborg Sealing Solutions, Vaihingen (Stuttgart)


    We offer you a challenging research topic, close supervision, and the opportunity to develop practical and theoretical skills. If you are interested, send your CV, transcript of records, and a brief letter of motivation to philipp.spitzer@kit.edu.