Custom parts for electromagnetic devices can be created on-demand and at a cheap cost using 3D printing. Such devices are extremely sensitive, and each component must be manufactured to exact specifications.
until recently, the only option to diagnose printing issues was to build, measure, and test a device or utilize in-line simulation,
both of which are very expensive and time-consuming.
To address this,
a Penn State-led research team developed a first-of-its-kind system for diagnosing printing problems in real-time using machine learning.
which was published in Additive Manufacturing, is described by the researchers as a key first step toward resolving 3D-printing problems in real-time.
According to the researchers,
this might significantly improve the efficiency of sensitive device printing in terms of time, cost, and quality.
Predicting Printing errors
“During the 3D printing process for any part, a lot of things may go wrong,” said Greg Huff,
an associate professor of electrical engineering at Penn State.
“And the electromagnetic world, where dimensions are measured in wavelengths rather than standard units of measurement,
even minor errors can lead to large-scale system failures or degraded operations.” If 3D printing a household object is like tuning a tuba — which can be done with broad adjustments — electromagnetic 3D-printing equipment is like tuning a violin: little tweaks count.”
The researchers added cameras to printer heads in a previous project, collecting an image every time something was produced.
Even though it was not the primary aim of the project, the scientists were able to compile a dataset that they could use with an algorithm to classify different sorts of printing problems.
“The heart of this research was generating the dataset and figuring out what information the neural network needed,” said first author Deanna Sessions,
who earned her doctorate in electrical engineering from Penn State in 2021 and now works as a contractor for the Air Force Research Laboratory for UES Inc.
“We’re leveraging this data—obtained from low-cost optical pictures — to predict electromagnetic performance without having to do simulations throughout
the manufacturing process.”
When the framework is applied to the print, it can detect mistakes as they occur during the printing process.
The potential of fixing faults during the printing process is much closer to being a reality
now that the electromagnetic performance impact of defects can be determined in real-time.
Source: Penn State