Fuzzy Clustering of Wavelet Features for Tool Condition Monitoring in High Speed Milling Process

Amin J. Torabi, Meng Joo Er, Xiang Li, Beng Siong Lim, Zhai Lianyin, Huang Sheng, San Linn, and Gan Oon Peen
Submission Type: 
Full Paper
Supporting Agencies (optional): 
Nanyang Technological University, SIMTech Singapore
phmc_10_015.pdf3.09 MBSeptember 3, 2010 - 2:40pm

Ball-nose end-milling process is considered as one of the vital metal removal processes for aerospace pieces. The product surface quality highly depends on how this process is performed. One of the factors that affect the resulting surface roughness is the performance of the ball-nose cutter used. However, this performance changes with the degradation of the tool. Therefore, several tool preparation methods are applied to increase tool life. Old tools seriously damage the surface. To prevent this, it is advised to use tools up to a certain level of damage. Principally, tool wear measurement needs tool extraction which is machine-time consuming. However, to make it feasible, side signals of cutting process are analyzed to be correlated with actual tool damage. These types of analysis facilitate non-intrusive tool condition detection which leads to more production time with non-stopping cutting process. Since the signals are continuous, their features are usually extracted for analysis. Clustering of these features leads to a common platform that makes it possible to generalize the results of some specific experimental cutting processes.
This paper focuses on fuzzy clustering of wavelet features to confirm the repeatability of the patterns in different cutters of the same batch. Some discussions on the clustering results will be provided.

Publication Control Number: 
Submission Keywords: 
signal processing
Fuzzy Clustering
Ball-Nose End-Milling
Wavelet Analysis
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed