Real-time process monitoring in additive manufacturing using machine learning

Additive manufacturing (AM) facilitates the production of complex, customized parts by building them up layer by layer, making it ideal for small batch production. However, the process is susceptible to defects and anomalies, particularly for low batch sizes, which can affect the quality of the final part. Given the dynamic nature of AM, it is possible to adjust process parameters in subsequent printing steps to mitigate the impact of detected anomalies. This paper proposes a two-step framework for real-time process monitoring and parameter adaptation, where anomalies are first detected and localized, followed by adaptive adjustments to the printing parameters. The effectiveness of the framework is demonstrated through a case study involving Fused Deposition Modeling.