Scholarlink

<p>This study proposes a machine learning based cutting tool wear predictive model using extracted features from real-time multi-sensor data. The real-time multi-sensor data including cutting tool vibration signals and surface texture data were collected during face milling operations. Both temporal and spatial information was preserved by transforming the tool vibration signal into polar coordinates using the Gramian angular field (GAF). Feature extraction from the sensor data is conducted using the Gabor wavelet transform (GWT), followed by principal component analysis (PCA) to identify significant features. Subsequently, a tool wear predictive model was established using various machine learning models including Random Forest (RF), K-Nearest Neibhor (KNN), and extreme gradient boosting (XGB). The XGBoost model outperformed the other two models with the lowest test MSE of 0.00100 and the highest R² of 0.95. Both Grid Search and Random Search yielded the same best score of -0.0011, with similar optimal parameters.</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

A machine learning algorithm based cutting tool wear assessment using multi-sensorial data

  • 期刊论文

  • 发表日期：

    2025年1月10日

查看全文
  • Mulpur Sarat Babu

摘要

This study proposes a machine learning based cutting tool wear predictive model using extracted features from real-time multi-sensor data. The real-time multi-sensor data including cutting tool vibration signals and surface texture data were collected during face milling operations. Both temporal and spatial information was preserved by transforming the tool vibration signal into polar coordinates using the Gramian angular field (GAF). Feature extraction from the sensor data is conducted using the Gabor wavelet transform (GWT), followed by principal component analysis (PCA) to identify significant features. Subsequently, a tool wear predictive model was established using various machine learning models including Random Forest (RF), K-Nearest Neibhor (KNN), and extreme gradient boosting (XGB). The XGBoost model outperformed the other two models with the lowest test MSE of 0.00100 and the highest R² of 0.95. Both Grid Search and Random Search yielded the same best score of -0.0011, with similar optimal parameters.