Scholarlink

<p>This study proposes a novel online monitoring system for chatter vibration detection in milling processes, utilizing the Variational Mode Decomposition (VMD) technique in conjunction with Spectral Entropy. VMD, an advanced and robust signal processing method, is employed to decompose audio signals captured during milling into Intrinsic Mode Functions (IMFs), enabling the analysis of complex signal patterns. Controlled milling experiments were conducted by varying key cutting parameters, and the signals were processed to extract 16 statistical features, including Root Mean Square (RMS), Waveform Index, Skewness Index, Crest Factor, Log Energy Entropy, Power Spectral Entropy, Spectral Entropy, and Log Detector. Among these features, Spectral Entropy demonstrated superior predictive capabilities in distinguishing chatter conditions—stable, transient, and unstable states. For classification, multiple machine learning models were applied, including Random Forest, J48 Decision Tree, and REP Tree classifiers. The J48 classifier achieved the highest accuracy of 86.11%, followed by the REP Tree classifier with 82.22%, and the Random Forest classifier with 81.11%. These results highlight the effectiveness of the J48 classifier in providing precise and reliable classification of chatter states. The proposed approach significantly contributes to the development of an effective real-time chatter vibration monitoring system, enhancing diagnostic accuracy and offering innovative solutions for improving machining operations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Measurement and Prediction of Chatter Stability in CNC Milling Using VMD and Decision Tree Classification

  • 期刊论文

  • 发表日期：

    2025年9月3日

查看全文
  • Pushpendra Kumar Kushwaha,
  • Pankaj Gupta,
  • Rohit Mishra

摘要

This study proposes a novel online monitoring system for chatter vibration detection in milling processes, utilizing the Variational Mode Decomposition (VMD) technique in conjunction with Spectral Entropy. VMD, an advanced and robust signal processing method, is employed to decompose audio signals captured during milling into Intrinsic Mode Functions (IMFs), enabling the analysis of complex signal patterns. Controlled milling experiments were conducted by varying key cutting parameters, and the signals were processed to extract 16 statistical features, including Root Mean Square (RMS), Waveform Index, Skewness Index, Crest Factor, Log Energy Entropy, Power Spectral Entropy, Spectral Entropy, and Log Detector. Among these features, Spectral Entropy demonstrated superior predictive capabilities in distinguishing chatter conditions—stable, transient, and unstable states. For classification, multiple machine learning models were applied, including Random Forest, J48 Decision Tree, and REP Tree classifiers. The J48 classifier achieved the highest accuracy of 86.11%, followed by the REP Tree classifier with 82.22%, and the Random Forest classifier with 81.11%. These results highlight the effectiveness of the J48 classifier in providing precise and reliable classification of chatter states. The proposed approach significantly contributes to the development of an effective real-time chatter vibration monitoring system, enhancing diagnostic accuracy and offering innovative solutions for improving machining operations.