Applications of machine learning in process development for metal 3D printing

• Main Author: Lam, Kenny Yijie
• Other Authors: Huang Changjin
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/153962
• Institution: Nanyang Technological University

Additive manufacturing (AM) became popular in the 1980s which involves printing objects layer by layer with the aid from a 3D Computer Aided Design (CAD) model. In mid 1990s, Machine Learning started to flourish as a means of using its applications like Artificial Intelligence to solve complex problems. Deep learning and Artificial Neural Networks are mainly used to identify underlying patterns in the data to optimize AM process parameters. However, a large dataset is required to accurately carry out Deep Learning Machine Learning techniques. This report on the other hand tackles the issue of using a small dataset to optimize AM process parameters using techniques to reduce model complexity and prevent over-fitting, while producing accurate predictions. This research helps to make printing of the relatively new FeCoNiCrAl(0.1) alloy using the Direct Energy Deposition process, a more efficient and cost-effective by making use of Machine Learning techniques. This project aims to carry out in-depth research into the optimization of process parameters using Machine Learning techniques. This project covers surface analysis and mechanical hardness analysis for Direct Energy Deposition of FeCoNiCrAl(0.1) alloy obtained experimentally. Using Machine Learning, process parameters will be analyzed with experimental outputs such as hardness, functions of height, width and melting pool depth. Machine Learning models are then trained to and testing to be used to find optimized process parameters.