Application of ML and AI to optimize hybrid 3D metal printing

This work aims to optimize the additive part of the process for Hybrid 3D metal printing. WMU created a hybrid 3D metal printer that integrates additive and Computer Numeric Control (CNC) machining processes. Metal layers are formed using gas metal arc welding (GMAW), with CNC controlling the deposition path. After a layer is deposited, the milling tool from the same CNC machine removes the excessive material. The process is repeated to complete the final part with tolerance and finish surfaces matching the standards of CNC machining.
While this hybrid manufacturing process has been tested to create good parts, some improvements can be made to increase the speed, reduce material, and allow operation with operator intervention.
Several optimization algorithms have been developed using AI and ML algorithms to improve the quality of parts produced with the hybrid 3D metal printer. First, we measure hundreds of samples from various parameter settings for additive deposition. To have a complete set of testing and verification, samples were measured manually with a digital gauging system, using a high-definition vision system for 2D measurement of layer profile and 3D high-resolution scanning. Collected data is processed with an ML algorithm and tested against the measured values. Second, layers with sizes from ML were calculated. Finally, an additive layer of the metal deployment following the path is created using an AI generic algorithm, and the CNC program is uploaded to the hybrid machine. During the 3D printing process, the vision system monitors and controls the additive and machining parameters. After completing the layer, the process is repeated. By correcting defects during printing, this system ensures consistently high-quality layers.
For industrial applications, this 3D metal printer offers automated control, AI/ML based, software optimization for part creation, and in-process vision/3D scanning quality inspection.