Machine Learning Method to Predict 3D Printing Accuracy Wins IEEE CASE 2021 Award in France – USC Viterbi

Cesar Ruiz, Qiang Huang, and Yuanxiang Wang received the Best Lecture Paper Award at the 2021 IEEE International Conference on Automation Science and Engineering. Image/Qiang Huang

A new model that dramatically improves 3D printing accuracy was recently honored with the IEEE Best Conference Paper Award at the 17and International Conference on Automation Science and Engineering (CASE), presented by the IEEE Robotics and Automation Society in Lyon, France.

The work was presented by Ph.D. candidate in the Department of Industrial and Systems Engineering Daniel J. Epstein, Yuanxiang Wang, along with postdoctoral researcher Cesar Ruiz and Professor of Industrial and Systems Engineering Qiang Huang.

3D printing, or additive manufacturing, has gained momentum as a new manufacturing technology that has the potential to revolutionize many industries. However, 3D printing is also prone to printing imperfections. Finished products may deviate from the design for a variety of reasons, such as complexity of product shape, print process control, and material properties. In order to print products successfully, shape prediction models often have to learn from a wide range of shape deviation data.

Wang and his collaborators have now developed an engineering-informed machine learning methodology based on small data. The model can learn and predict deviations of smooth and non-smooth 3D shapes in a unified modeling framework.

“Shape deviation patterns vary depending on shape geometries, sizes, materials, and 3D printing processes,” Wang said. “Due to customization and high manufacturing costs, only a small number of samples are printed, each with a very distinct shape deviation pattern. It is extremely difficult to learn and predict geometric quality with a small, heterogeneously shaped dataset.

Wang said that by learning heterogeneous shape deviation data, the model can establish the association between smooth baseline shapes and non-smooth shapes.

The new machine learning model can learn from heterogeneous data to predict shapes of smooth and non-smooth objects.

The new machine learning model can use data to learn and predict how the shape of a printed object may deviate.

“Essentially, we break down the additive manufacturing process mathematically into two steps: additively fabricate smooth basic shapes using a convolution framework to model layer-to-layer interactions, then remove additional materials at the same time. using a cookie cutter function,” Wang said.

The new machine learning framework provides a new data analysis tool for shape engineering in additive manufacturing and beyond.

“The unified modeling approach we developed is able to predict the distortion of various geometries by enabling knowledge transfer from smooth to non-smooth shapes,” Wang said. “We are in the process of extending this methodology to freeform 3D shapes.”

Wang said the new model can be applied to broader engineering fields beyond 3D printing. Under Huang’s supervision, Wang and his teammates are also developing prototype software to demonstrate the methodologies for broader impacts. Visit the Huang Lab website for more information.

Posted on September 13, 2021

Last updated September 13, 2021

Comments are closed.