Multiworking Conditions Anomaly Detection of Mechanical System Based on Conditional Variational Auto-Encoder

Table 7

Anomaly detection accuracy of different methods.

Dataset	Baseline method
	PCA		LOF		VAE		CVAE		MW-CVAE
	F1-score	AUC	F1-score	AUC	F1-score	AUC	F1-score	AUC	F1-score	AUC
CWRU	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.01.01.0	1.01.01.0
JNU	0.8464	0.8986	0.8484	0.8981	0.7448	0.6949	0.8155	0.8895	0.98190.99220.9829	0.99730.99790.9970
PU	0.6606	0.6723	0.6441	0.5091	0.6604	0.6109	0.6441	0.4972	0.76530.69980.7185	0.63880.39870.6075

The bold type indicates that the method obtains the best performance index in this database. For example, in database PU, 0.6723 (bold) indicates that PCA methods outperform other methods on AUC performance indicators. In addition, MW-CVAE wins every performance index (F1-score and AUC).