An unsupervised feature learning based health indicator construction method for performance assessment of machines

•A multi-scale CAE network is used to extract features from three scale levels.•Only the data collected under health conditions are used for network updating.•A weighted relative similarity is proposed to construct health indicators.•Bearing and milling datasets are used to verify the performance of...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	Mechanical systems and signal processing Ročník 167; s. 108573
Hlavní autoři:	Guo, Liang, Yu, Yaoxiang, Duan, Andongzhe, Gao, Hongli, Zhang, Jiangquan
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	Berlin Elsevier Ltd 15.03.2022 Elsevier BV
Témata:	Construction methods Data acquisition Degradation Domains Feature extraction Genetic algorithm Genetic algorithms Health indicator Machine learning Multiscale convolutional autoencoder network Performance assessment Prognostics Related correlation distance Similarity Multiscale convolutional autoencoder network Related correlation distance Genetic algorithm Prognostics Health indicator
ISSN:	0888-3270, 1096-1216
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Buďte první, kdo okomentuje tento záznam!