Artificial intelligence: way forward to empower metal additive manufacturing product development – an overview

Abstract

Metal additive manufacturing (MAM) has emerged as a promising technology to fabricate parts that are nearly impossible to do so by traditional methods. Since MAM is considered an open-loop system, it is not as efficient at fabricating repeatable and high-quality parts as a closed system. Hence, changing the MAM open-loop system to a closed system proves to be an imperative step for producing reliable parts. The direct application of physics or heat transfer or fluid dynamics equations to a MAM system is impossible as several uncontrolled parameters impact each layer fabrication. Thus, leading researchers are focusing on developing a digital twin, a real-time virtual model or digital representation to provide better control of the MAM process. Currently, the digital twin for MAM is in the development stage as the data capture through real-time monitoring, experimental values, and simulations from layer fabrication need to be sorted and stored in a structured manner. The way forward for a consistent quality MAM part is achieved through a digital twin which is explicitly driven by subsets of artificial intelligence such as machine learning and deep learning. More data must be further acquired and utilized to train the machine to learn the algorithm. An overview of artificial intelligence in additive manufacturing for a closed-loop system is presented.