About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
2023 Technical Division Student Poster Contest
|
Presentation Title |
SPU-6: Assessing Laser Powder Bed Additive Manufacturing Part Quality via In-Situ Monitoring & Machine Learning |
Author(s) |
Ana Shukri Love |
On-Site Speaker (Planned) |
Ana Shukri Love |
Abstract Scope |
In-situ monitoring paired with real-time data analytics has the potential to overcome challenges associated with ensuring build quality and consistency in additive manufacturing (AM). Internal flaws are expensive and time consuming to identify in a completed part, however, in-situ monitoring enables inexpensive quantification of the size and spatial distribution of defects. This work focuses on the development of a framework that feeds real-time sensor data into machine learning algorithms trained to discriminate between nominal and defective build regions. Optical imaging and analysis performed using Oak Ridge National Laboratories’ Peregrine software is used to identify spatter and balling defects in real-time. Addition of thermal and acoustic sensors enables cross-correlation between multiple data streams. Validity of in-situ analysis for AM builds is determined by testing the neural network against user-specified ground truths. Ultimately, the use of functional and accurate in-situ monitoring will increase trust in AM, expanding its use in mainstream production.
|
Proceedings Inclusion? |
Undecided |
Keywords |
Additive Manufacturing, Machine Learning, |