11. Quality Prediction for Welding


Diana Peters

The Digital Production Platforms Group at the Institute of Data Science of the German Aerospace Center (DLR) is concerned with automated data exchange along the production process.

The main aspect of our research is how to exchange data between different stations and parties along the lifecycle of a product – automatically and maschine readable.

This includes semantic modelling and NLP – the machine learning expertise for this project will be provided by our ML group.

What is the data science project’s research question? For this project, sensor data (e.g. forces and vibrations) from a welding process is provided for different pieces. The same pieces were later tested for quality. The research question for this project is, how the quality of the final product can be predicted by the sensor data from the process.

What data will be worked on? Sensor data (time series and input variables) from a welding process (ultrasonic welding) and quality characterizations (failure mechanism and failure loads (measured values).

What tasks will this project involve?

Get a rough overview of the state of the art and the problem domain (very rough – the supervisors will provide some input for this) and understand the provided data.

Select (an) ML method(s) to analyze the data and find connections between the sensor data and the quality categorization.

Implement the selected method(s) and test them on the provided data; ensure scalability of the implementation.

What makes this project interesting to work on? In this project the scientist can get a first overview of production processes and IoT as well as the applicability of machine learning methods in this context.

What is the expected outcome? Contribution to research paper, Contribution to software development

What infrastructure, programs and tools will be used? Can they be used remotely?  In the project, our GitLab will be used for development and also to exchange the data – this is accessible remotely. We will also use this to start writing at the paper together. Other tools, like an IDE for Python, an editor for LaTex, etc., can be used locally – the infrastructure to exchange everything is our GitLab.

What skills are necessary for this project? Data analytics / statistics, Scientific computation, Data mining / Machine learning, experience with Python, preferably with Tensorflow, Keras, or PyTorch; also, Computational models, Deep Learning, and/or Visualisation would be helpful additional skills, but they are not necessary

Is the data open source? no 

Interested candidates should be at Phd level . Diana Peters is looking for 1 visiting scientist, working on the project together with the team.