Randomized trial demonstrates enhanced process optimization in aerospace manufacturing, indicating significant advancements in precision and material integrity.
Key Points
This study aims to optimize the Directed Energy Deposition process for aerospace components using machine learning techniques.
Proposed a hybrid optimization framework combining traditional experimental design with machine learning.
Employed the Taguchi method for coarse screening of control factors based on L16 orthogonal array.
Used a Fully Connected Neural Network with Bayesian Optimization to guide the optimization process.
Achieved a porosity as low as 0.03% in the optimized process.
Demonstrated an average tensile strength of approximately 1358 MPa and a hardness of ~40 HRC.
Validated the framework through robotic DED fabrication of a combustion chamber casing.