| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
| Presentation Title |
Leveraging Convolutional Neural Networks for the Prediction of Enhanced Plume and Coating Quality in Atmospheric Plasma Spraying |
| Author(s) |
Giuseppe A. Bianco Atria, Abhijith Sukumaran, Cheng Zhang, Arvind Agarwal |
| On-Site Speaker (Planned) |
Giuseppe A. Bianco Atria |
| Abstract Scope |
Atmospheric plasma spraying (APS) is a method used for producing layer-by-layer protective coatings that often necessitates a time-consuming trial-and-error approach due to the intricate correlation of various operational parameters like carrier gas pressure, amperage, and powder characteristics. To address this, we've applied Convolutional Neural Networks (CNN) to develop a predictive model that uses in-flight sensor data to capture spatial variations of particle parameters within the plasma plume relative to these inputs. Additionally, we've designed a plume score system based on output parameters such as temperature, particle velocity, flow rate, and diameter. We've identified the plume conditions that yield the best coating quality using an optimization process. By producing and characterizing coatings under these "best" and "worst" plume conditions, we've been able to compare their performance, thereby demonstrating the potential of our CNN-based approach in improving APS process control, efficiency, and effectiveness. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Computational Materials Science & Engineering, Process Technology |