Applications of machine learning in metal powder-bed fusion in-process monitoring and control: status and challenges

The continuous development of metal additive manufacturing (AM) promises the flexible and customized production, spurring AM research towards end-use part fabrication rather than prototyping, but inability to well control process defects and variability has precluded the widespread applications of AM. To solve these issues, process monitoring and control is a powerful approach. Recently, a variety of monitoring methods have been proposed and integrated with metal AM machines, which enables a large volume of data to be collected during the process. However, the data analytics faces great challenges due to the complexity of the process, bringing difficulties on developing effective models for defects detection as well as feedback control to improve quality. To overcome these challenges, machine learning methods have been frequently employed in the model development. By using machine learning methods, the models can be built based on the collected dataset, while it is not necessary to fully understand the process. This paper reviews the applications of machine learning methods in metal powder-bed fusion process monitoring and control, illuminates the challenges to be solved, and outlooks possible solutions.

Automated summary

The continuous development of metal additive manufacturing promises flexible and customized production, but the inability to control process defects and variability has hindered its widespread applications. Process monitoring and control, utilizing machine learning methods, can overcome these challenges and improve quality. This paper reviews the applications of machine learning in metal powder-bed fusion process monitoring and control, highlighting the challenges and possible solutions.