Incremental learning of LSTM-autoencoder anomaly detection in three-axis CNC machines

Li, Eugene; Li, Yang; Bedi, Sanjeev; Melek, William; Gray, Paul

doi:10.1007/s00170-023-12713-2

Cited by 3 publications

(2 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although the resulting system does not have the same level of accuracy as the system trained from scratch, the experiments show that this approach is feasible. A potential option would be to improve the output of the transfer learning system through using techniques like incremental ensemble learning (Li et al (2023b)), but this approach is outside the scope of this work.…”

Section: Discussionmentioning

confidence: 99%

“…Autoencoder are powerful tools that can be used for many purposes. In our previous work, we were able to successfully use LSTM-Autoencoders for chatter detection in CNC machines (Li et al (2023a), Li et al (2023b)). This is possible because Autoencoder are capable of capturing the embedded representation of a system with the encoder portions of the network.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Multi-class anomaly classification for CNC milling machines

Li,

Bedi,

Melek

2024

Preprint

View full text Add to dashboard Cite

Anomaly detection in CNC milling machine operation is a necessary component to develop a smart, autonomous CNC milling machine. These anomalies range from phenomena such as chatter, to tool breakage, to work piece misalignment. There have been many methods for detecting different anomalies, but many of these methods are focused on classifying one anomaly at a time. In this work, we propose an LSTM-Autoencoder based system to apply dimensionality reduction on accelerometer signals, and then apply Linear Support Vector Classifier (SVC) for multi-class classification and prediction.We extend this work to Autoencoders trained by transfer learning to reduce the data burden and validate both methods experimentally.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%