STAR Scholar 2019
In-situ Monitoring for FDM Printers with Acoustic Emission
Fused deposition modeling (FDM) is the mainstream additive manufacturing process available for consumers. This process deposits sequential layers of polymer in layered profiles to achieve complex geometries with a single machining setup. With this robust manufacturing process comes disadvantages that include poor reliability and inconsistent part production. To address these drawbacks, acoustic emission (AE) technology was employed to monitor a 3D printer’s states during the manufacturing process. Specifically, this project focused on developing an effective closed-loop system that provides sensory feedback to monitor FDM machine states. To achieve this goal, AE data collected in real-time during printing was first processed offline using a machine learning method capable of attributing sets of data to a given machine state. A database of AE data was then developed for critical states on a given 3D printer. The concept was then validated in real-time by programming a controller to receive AE data and running the machine learning algorithm to diagnose changing states of the printer. Future steps for this project include the integration of the identified states with the actual printing process to enable higher-quality printing.
In-situ Monitoring for FDM Printers with Acoustic Emission
Fused deposition modeling (FDM) is the mainstream additive manufacturing process available for consumers. This process deposits sequential layers of polymer in layered profiles to achieve complex geometries with a single machining setup. With this robust manufacturing process comes disadvantages that include poor reliability and inconsistent part production. To address these drawbacks, acoustic emission (AE) technology was employed to monitor a 3D printer’s states during the manufacturing process. Specifically, this project focused on developing an effective closed-loop system that provides sensory feedback to monitor FDM machine states. To achieve this goal, AE data collected in real-time during printing was first processed offline using a machine learning method capable of attributing sets of data to a given machine state. A database of AE data was then developed for critical states on a given 3D printer. The concept was then validated in real-time by programming a controller to receive AE data and running the machine learning algorithm to diagnose changing states of the printer. Future steps for this project include the integration of the identified states with the actual printing process to enable higher-quality printing.