Scholarlink

<p>This research project developing an real time tool wear monitoring system for CNC turning operations, leveraging a computer vision and predictive modelling to enabling a real-time tool condition assessment. This system addresses critical industry challenges of unplanned downtime, tooling costs, and quality control due to uneven tool wear, through an innovative non-contact approach that extracts machining parameters directly from CNC displays. The core methodology integrates optical character recognition (OCR) using OpenCV and Tesseract with a physics-based wear model adapted from Choudary’s equations. A calibrated camera captures spindle load, speed, and feed rate data at regular intervals, achieving 98.5% parameter extraction accuracy under typical shopfloor lighting conditions. This model processes these inputs to predict flank wear progression with ± 0.05&#xa0;mm precision for wear under 0.6&#xa0;mm. Experimental validation involved 32 controlled turning trials on SS304 workpieces using CNMG120408 inserts. Results demonstrated strong correlation (R² = 0.94) between predicted and measured the wear values, with surface roughness measurements providing secondary validation (R² = 0.89). This system reduced scrap rates by 66% through early wear detection while the extending tool life by 18% via optimized replacement timing. Key advantages include non-invasive implementation requiring no machine modifications and low-cost on deployment whether it costs 15,000 + for conventional sensor systems. Current limitations involve the performance under extreme loads (&gt; 85% spindle capacity) and dependency on Fanuc control displays, with solutions for other CNC brands under development.</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

Tool wear prediction system in single point cutting tool

  • 期刊论文

  • 发表日期：

    2026年5月14日

查看全文
  • P. Karuppuswamy,
  • Ashwanthkanna K,
  • Rangesh Prabhu B,
  • Sridhar S

摘要

This research project developing an real time tool wear monitoring system for CNC turning operations, leveraging a computer vision and predictive modelling to enabling a real-time tool condition assessment. This system addresses critical industry challenges of unplanned downtime, tooling costs, and quality control due to uneven tool wear, through an innovative non-contact approach that extracts machining parameters directly from CNC displays. The core methodology integrates optical character recognition (OCR) using OpenCV and Tesseract with a physics-based wear model adapted from Choudary’s equations. A calibrated camera captures spindle load, speed, and feed rate data at regular intervals, achieving 98.5% parameter extraction accuracy under typical shopfloor lighting conditions. This model processes these inputs to predict flank wear progression with ± 0.05 mm precision for wear under 0.6 mm. Experimental validation involved 32 controlled turning trials on SS304 workpieces using CNMG120408 inserts. Results demonstrated strong correlation (R² = 0.94) between predicted and measured the wear values, with surface roughness measurements providing secondary validation (R² = 0.89). This system reduced scrap rates by 66% through early wear detection while the extending tool life by 18% via optimized replacement timing. Key advantages include non-invasive implementation requiring no machine modifications and low-cost on deployment whether it costs 15,000 + for conventional sensor systems. Current limitations involve the performance under extreme loads (> 85% spindle capacity) and dependency on Fanuc control displays, with solutions for other CNC brands under development.