Digitalization of production process for additive manufacturing - image segmentation and optimization

Purpose and goal:

Additive manufacturing allows for the development of components with unique properties and complex geometries. The process is complex and involves several printing parameters to achieve the desired microstructures. A bottleneck with the method is the time-consuming analysis of the properties of the materials produced.

This includes microscopy and image analysis. The aim of this project is to construct AI algorithms that can fully automate this process, enabling a fast and efficient characterization of microstructural properties in AM materials. Expected results and effects: ** Denna text är maskinöversatt **

The results of this project are expected to provide new methods for controlling the quality of additively manufactured materials. This will enable an increase in productivity and reduce the amount of scrap from the production process. A fully automated characterization method will help experts spend fewer hours on analysis, which accelerates the development of additively manufactured materials in a sustainable way.

Approach and implementation:

Machine learning and deep learning methods will be used to make libraries of trained models that can segment metallographic images, classify and present quantitative results. Other set of trained algorithms will be made when complementary characterization data from other methods, for example diffraction and or mechanical properties, are available. These will correlate the results and provide overall information about the prints and can be used to predict lacking data.

A feasibility test will be done to find a correlation between printing parameter and the quality of final prints.