Real-time FDM machine condition monitoring and diagnosis based on acoustic emission and hidden semi-Markov model

Machine condition monitoring is considered as an important diagnostic and maintenance strategy to ensure product quality and reduce manufacturing cost. However, currently, most additive manufacturing (AM) machines are not equipped with sensors for system monitoring. In this paper, a real-time lightweight AM machine condition monitoring approach is proposed, where acoustic emission (AE) sensor is used. In the proposed method, the original AE waveform signals are first simplified as AE hits, and then segmental and principal component analyses are applied to further reduce the data size and computational cost. From AE hits, the hidden semi-Markov model (HSMM) is applied to identify the machine states, including both normal and abnormal ones. Experimental studies on fused deposition modeling (FDM), one of the most popular AM technology, show that the typical machine failures can be identified in a real-time manner. This monitoring method can serve as a diagnostic tool for FDM machines.