Method for monitoring of FDM 3D printer failure based on acoustic emission

Authors：Wu Hai Xi; Yu Zhong Hua^*; Zhang Hao; Yang Zhen Sheng; Wang Yan

Source：Zhejiang Daxue Xuebao (Gongxue Ban)/Journal of Zhejiang University (Engineering Science), 2016, 50(1).

DOI：10.3785/j.issn.1008-973X.2016.01.012

• Clustering results
• Fused deposition modeling
• K-Means clustering algorithm
• Monitoring methods
• Real time monitoring
• Relevant features
• Time resolution
• Typical failures

A monitoring method based on acoustic emission (AE) was proposed aiming at the typical failure modes of material filament breakage or run out and extruder blockage in the extruder of fused deposition modeling (FDM) 3D printer. Two experiments were designed and conducted accordingly. The AE signals were processed and the related features were extracted parametrically based on AE hits in order to reduce the costs on sensor data computing and storing and improve the real-time monitoring performance. Sensor data from the experiments were collected and analyzed. The relationship between the features of AE hits and failure modes was estimated. The knowledge of the most relevant features of AE hits was obtained. The K-means clustering algorithm was applied to simultaneously identify the two types of failure modes based on the AE features of absolute energy and counts respectively. Clustering results of the proposed monitoring method showed that the accuracy rates were 94.62% and 93.80% under the time resolution of 0.2 s.

• Institution
  Shanghai Maritime University; Zhejiang University; georgia institute of technology