Please use this identifier to cite or link to this item:
Title: Effective application of neural networks in tool condition identification
Authors: Sun, J. 
Rahman, M.
Wong, Y.S. 
Hong, G.S. 
Sun, Y.
Keywords: Hybrid system
Neural networks
Tool condition monitoring
Tool wear
Issue Date: 2005
Citation: Sun, J.,Rahman, M.,Wong, Y.S.,Hong, G.S.,Sun, Y. (2005). Effective application of neural networks in tool condition identification. LEM 2005 - 3rd International Conference on Leading Edge Manufacturing in 21st Century : 893-898. ScholarBank@NUS Repository.
Abstract: Neural networks (NNs) are widely used in intelligent tool condition monitoring (TCM) system, however, its effective application is still an ongoing research issue. This paper investigates some problems related with this topic, such as the selection of monitoring signals and signal feature set. and construction of performance evaluation function in TCM decision making, so as to realize reliable tool condition identification. Firstly, sensing signal selection and feature set selection are investigated. The former aims to choose the proper monitoring signals in terms of the above mentioned requirements. And the latter is to comprehensively take this monitoring signal features and identify the most effective set so as to give robust and reliable tool condition identification. From the viewpoint of reducing manufacturing cost, an evaluation function with the machining loss consideration is proposed as criteria in decision making. Besides, two related problems in TCM system design are discussed, multiclassification of tool states and hybrid system. Acoustic emission (AE) signals and cutting force signals from titanium machining, and AE signals from steel machining are used in this study.
Source Title: LEM 2005 - 3rd International Conference on Leading Edge Manufacturing in 21st Century
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

Page view(s)

checked on Apr 12, 2021

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.