In Situ Quality Monitoring in AM Using Acoustic Emission: A Machine Learning Approach

Most control methods in additive manufacturing (AM) are mainly based on temperature or high resolution imaging. However, no methods are known to monitor the quality of AM in realtime. Our approach is very innovative for the quality monitoring of the process online by combining acoustic emission (AE) with machine learning. The machine parameters were changed when processing a steel 316L to get three quality levels for AE data acquisition, processing and validation. We demonstrate that the AM process has a number of unique acoustic signatures that can be detected and interpreted in terms of quality. The classification of AE is made by machine learning (ML) methods. This includes the extraction and recognition of unique acoustic signatures from various sintering or melting events. The confidence level achieved in the classification is very high showing that our approach is very promising for in situ and real-time monitoring of AM processes.