User login


You are here

Postdoc (2 years) on Machine learning techniques for interpretation of vibrational data

Additive manufacturing of metal alloys yields great potential for the aerospace industry (and others) as it allows the generation of geometrically complex structures with high specific strength, low density and high corrosion resistance. For example, General Electric has demonstrated the possibility of printing titanium fuel injectors for their LEAP engine, Boeing incorporated more than 300 printed parts in their 777X airplane … For such critical applications, the structural quality of printed parts is of utmost importance. Small deviations in print conditions, e.g. change in laser fluence, variation in powder quality, non-uniform gas flow, build position … may result in varying print quality, and even to inferior parts containing porosity, cracks, residual stress and geometrical deviations, amongst others. Current inspection protocol involves the use of X-rays, but this is a time-and cost-inefficient method for high-volume applications.

The RESON-AM project aims at developing a fast, practical and sensitive quality control technologies for AM metal parts. The project is a close collaboration between several large industries (Materialise, Siemens and MatchID) and academia (KULeuven and UGent). Materialise will provide large sets of AM metal parts for use in the project. The project is funded by SIM-Flanders (Strategic Initiative Materials in Flanders).

At UGent, three researchers will be involved in the project who will focus on the development and implementation of a quality inspection strategy of AM parts using vibrations. For this specific vacancy, we search a postdoctoral researcher who will work on data-driven models for simulating the vibrational behavior of AM metal parts, and for generating a large and diverse virtual database to enrich the experimental database. Secondly, the researcher will investigate and implement (supervised) classification methods to identify long-term trends in the printed parts (e.g. due to drift of laser fluence), and to detect off-nominal AM parts (e.g. due to presence of large pores) using the virtual and experimental database.

This vacancy applies to mature researchers having a strong background in statistics and deep learning. Knowledge in vibrational testing is a plus.

More information can be found on:

Subscribe to Comments for "Postdoc (2 years) on Machine learning techniques for interpretation of vibrational data"

Recent comments

More comments


Subscribe to Syndicate