A Machine Learning trick that everyone should be using

00:00 Introduction
01:18 A machine learning solution in 3D Printing
03:29 Adding physics to machine learning
05:46 Paper, news, and blog
06:53 Impact
07:47 Other related videos

In some cases, machine learning can be used with a small data set because the behavior of many complex engineering systems is often accurately described by a group of variables rather than individual variables. An example is the well-studied problem of the flow of fluid in a pipe. In principle, four variables, i.e., the pipe diameter, average fluid velocity, and the density and viscosity of the fluid can predict if the flow is laminar or turbulent. However, it is well accepted that the nature of the flow, laminar or turbulent, can be predicted by only one variable, the Reynolds number. The reduction in the number of variables can make many problems tractable. We provide an example of preventing cracks in additively manufactured metallic parts. Please watch the video, read the papers and/or review the news item in MRS Bulletin and/or read the blog in GitHub.

1) B. Mondal, T. Mukherjee, T. DebRoy. Crack-free metal printing using physics-informed machine learning. Acta Materialia. 2022, Vol. 226, Article # 117612. Available at: https://www.sciencedirect.com/science...
2) Y. Du, T. Mukherjee, T. DebRoy. Physics-informed machine learning and mechanistic modeling of additive manufacturing to reduce defects. Applied Materials Today. 2021, Vol. 24, Article # 101123. Available at: https://www.sciencedirect.com/science...
3) MRS (Materials Research Society) Bulletin News: https://link.springer.com/article/10....
4) GitHub blog: https://jigarp12892.github.io/posts/2...

#additivemanufacturing #3dprinting #machinelearning #cracking #modeling #physics