Acoustic process monitoring for selective laser melting (SLM) with neural networks: A proof of concept

Selective laser beam melting (LBM) is currently on the verge of being used for small series production. A major disadvantage of the process is currently the low re-producibility of the process quality. Some current research work therefore concentrates on the integration of optical measurement technology for process monitoring. In addition to optical methods, initial investigations show that acoustic sensors for process monitoring are also a promising approach. Data processing poses a great challenge for acoustic data, as the raw acoustic signal is difficult to interpret. In this paper a new concept for an acoustic process monitoring is presented and integrated into a test environment. In order to record acoustic signals, process parameters are varied in a Design of Experiments and test objects of different quality are built up. An artificial neural network is trained to evaluate the used process parameters (three laser powers) for a first proof of the concept of the measuring system for the monitoring of the process. This work shows that this measuring technique has the potential to monitor the process.