Machine learning in solid state additive manufacturing: state-of-the-art and future perspectives

Machine learning (ML) has recently found compelling applications in the manufacturing and materials industries. With this state-of-the-art technology, materials innovation, critical simulations for product manufacture, and predictive insights that can significantly improve overall production efficiency are all being accomplished. Implementing a robust ML model promises to minimise the need for extensive physical trials, resulting in substantial time and cost savings. This article overviews the applications of ML to several aspects of the solid-state additive manufacturing (SSAM) spectrum, including response prediction methods, design for SSAM, process parameter prediction, in-situ monitoring, quantitative microstructural analysis, grain size distribution, critical velocity analysis with feature selection, surface roughness prediction, and post-processing etc. The existing difficulties in implementing ML in SSAM and prospective solutions are discussed in detail. Furthermore, upcoming trends and solutions are proposed to offer a comprehensive overview of the domain of SSAM. ML models have decreased cost and time-consuming experiments, making the process more efficient and cost-effective. The complexity of the manufacturing process and the lack of relevant data hinder ML deployment in SSAM. Encouragingly, efforts are underway to solve these problems and improve machine learning in solid-state additive manufacturing.