Machine learning in additive manufacturing: A comprehensive insight

Additive manufacturing (AM) is a technological advancement gaining colossal popularity due to its advantages and simplified fabrication. AM facilitates the manufacturing of complex, light, and strong products from digitized designs. With recent advancements, AM can bring digital flexibility and improved efficiency to industrial operations. Despite the various advantages, there is continuous variation in the qualities of AM products, which remains the main challenge in the wide application of AM. The performance of printed parts is directly influenced by processing parameters, and adjusting the parameters in the AM process can be quite challenging. The barrier can be minimized by proper monitoring of the AM process and precise measurement of AM materials and components, which is difficult to achieve through analytical and numerical models. Current research demonstrates machine learning (ML) and its techniques as a novel way to reduce costs. It also helps achieve optimal process design and part quality using the fundamentals of the AM process. ML is a subcategory of artificial intelligence (AI) that enables systems to learn and improve from measured data and past experiences. The present paper is focused on presenting a broad understanding of the current applications of ML in AM and thus provides a solid background for practitioners and researchers to apply ML in AM. Very few earlier reviews were presented before, but their studies mostly focus on artificial neural network technology and other irrelevant papers. In addition, most papers were published in 2021 and 2022 and were not discussed in earlier reviews. This state-of-the-art review is based on the latest database collected from Web of Science (WoS), Publons, Scopus, and Google Scholar using machine learning and additive manufacturing as the keywords. Extensive information collected on the possible applications of ML in AM shows that ML can be effectively applied to improve AM part quality and process reliability.