Intelligent analysis of tool wear state using stacked denoising autoencoder with online sequential-extreme learning machine

•An efficient OS-ELM algorithm is proposed to identify tool wear state.•A SDAE neural network is designed to reduce the samples dimension.•Spindle current signals are collected and analyzed.•The average testing accuracy and classification effect are significantly improved. Impeller is a critical com...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Measurement : journal of the International Measurement Confederation Jg. 167; S. 108153
Hauptverfasser: Ou, Jiayu, Li, Hongkun, Huang, Gangjin, Yang, Guowei
Format: Journal Article
Sprache:Englisch
Veröffentlicht: London Elsevier Ltd 01.01.2021
Elsevier Science Ltd
Schlagworte:
Artificial neural networks
Critical components
Cutting tools
Engineering
Hydroelectric power
Impellers
Machine learning
Machine tools
Machinery
Milling (machining)
Neural networks
Noise reduction
Online sequential extreme learning machine
Service life
Spindle current signals
Stacked denoising autoencoder
Tool wear
Tool wear states recognition
Tool wear states recognition
Online sequential extreme learning machine
Spindle current signals
Stacked denoising autoencoder
ISSN:0263-2241, 1873-412X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:•An efficient OS-ELM algorithm is proposed to identify tool wear state.•A SDAE neural network is designed to reduce the samples dimension.•Spindle current signals are collected and analyzed.•The average testing accuracy and classification effect are significantly improved. Impeller is a critical component of much large equipment, such as hydropower, nuclear pump, and so on. Its processing quality seriously affects the operation states and service life span of the equipment, while the milling cutter wear determines the processing accuracy of the impeller. Therefore, it is of great important to investigate an intelligent recognition method of tool wear state during the processing. In this research, a new method named stacked denoising autoencoder (SDAE) with online sequential extreme learning machine (OS-ELM) is put forward for intelligent recognition of tool wear states. Firstly, three current signals of the spindle of CNC machine tool are collected during the cutting process and the effective values are synthesized. Then, a new SDAE neural network is trained to acquire the low dimensional features using raw current signals. Finally, OS-ELM is used to realize recognition and classification of the milling cutter, and compared the accuracy with other methods. The method proposed in this paper was verified by both the laboratory signals and the actual engineering signals. The results for spindle current signals show that the developed model has achieved effective performance on tool wear recognition.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0263-2241
1873-412X
DOI:10.1016/j.measurement.2020.108153