|
CSME 2025/12
Volume 46 No.6
:
601-608
|
|
Spindle Imbalance Detection Method Based on Principal Component Analysis and Self-Organizing Map for Computer Numerical Control Machine Tools
Hao Yu Lin a, Ping Chun Tsai a, Ping Cheng Lin a and Jui Peng Hsu b
aDepartment of Mechanical Design Engineering, National Formosa University, Yunlin County, Taiwan (R.O.C.)
bPrecision Machinery Research Development Center, Taichung, Taiwan (R.O.C.)
|
Abstract:
This paper introduces a spindle imbalance detection method based on principal component analysis and a self-organizing map for machine tools under computer numerical control. By analyzing vibration signals collected during spindle operation, the method extracted eight key frequency-domain features, including spindle rotation and bearing characteristic frequencies. These features were refined through principal component analysis to eliminate collinearity while preserving essential information. A diagnostic model was established using a self-organizing map trained exclusively with healthy state data, enabling the autonomous monitoring of spindle health conditions. The method’s effectiveness was validated through extensive experiments on a YCM_NDV102A vertical machining center. Experiments revealed accuracy of 99.7% and 100% in identifying normal spindle conditions and imbalance states, respectively. Overall, the proposed method performed similarly to a supervised learning method but did not require fault data for model training, making it more suitable for industrial applications.
|
|
Keywords: spindle monitoring, spindle imbalance, principal component analysis, self-organizing map
|
Download PDF
|
©
2025
CSME , ISSN 0257-9731
|