Related Experiment Video
Updated: Nov 10, 2025
06:08
Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging
Published on: May 5, 2011
17.0K
Automatic Identification of Tool Wear Based on Thermography and a Convolutional Neural Network during the Turning
Nika Brili1, Mirko Ficko1, Simon Klančnik1
1Faculty of Mechanical Engineering, University of Maribor, Smetanova ul. 17, 2000 Maribor, Slovenia.
Sensors (Basel, Switzerland)
|April 3, 2021
Summary
This study introduces an automated system using infrared imaging and a convolutional neural network (CNN) to detect cutting tool wear during machining. The system accurately identifies tool conditions, improving manufacturing efficiency and safety.
Area of Science:
- Manufacturing Engineering
- Artificial Intelligence
- Materials Science
Background:
- Automated monitoring of cutting tool condition is crucial for efficient and safe machining operations.
- Traditional methods for tool wear detection often rely on subjective operator assessment or indirect measurements, leading to inconsistencies.
- Real-time, accurate detection of tool wear and damage is essential to prevent costly equipment failures and ensure product quality.
Purpose of the Study:
- To develop and validate a novel control system for automatic cutting tool condition supervision during turning operations.
- To leverage infrared thermography and machine learning for precise identification of tool wear levels.
- To enhance machining process reliability by enabling immediate response to tool degradation.
Main Methods:
- Utilized an infrared camera for capturing thermographic and visual data of the cutting process under controlled dry machining conditions.
- Collected a dataset of over 9000 images from machining low alloy carbon steel with tool inserts at various wear stages.
- Developed a convolutional neural network (CNN) model to classify cutting tool states (none, low, medium, high wear) based on thermographic data.
Main Results:
- Achieved a classification accuracy of 99.55% for predicting cutting tool wear and damage.
- Demonstrated the effectiveness of the CNN model in automatically determining the tool's condition using thermographic process data.
- Successfully implemented a system capable of close-up observation of machining despite challenging environmental factors like hot chips.
Conclusions:
- The proposed system provides a highly accurate and automated solution for cutting tool condition monitoring.
- Infrared thermography combined with CNNs offers a robust method for real-time tool wear assessment in manufacturing.
- This technology can significantly improve operational safety and efficiency by enabling prompt intervention, irrespective of operator expertise.
