Scholarlink

<p>This work presents an advanced metrology method for defect detection in rotating machine components using infrared thermography integrated modified artificial neural network (MANN) based image processing in a real-time manner. The proposed method utilizes the precise, accurate metrological features of infrared thermography, which are highly useful in accurate defect detection. Infrared thermography images are pre-processed by converting them to greyscale images, and then a median filter is used to remove the noise from the images. Thereafter, the proposed MANN method is used via MATLAB program for fault detection and finding the accurate location of the fault. The quantitative analysis is performed on synthetic star images using performance metrics like accuracy, Jaccard similarity index, Dice similarity index, Sensitivity, and Precision. The resulting yield of the proposed method is measured and compared with the yield of various benchmark edge-based segmentation methods from the literature, and it is found that the proposed MANN method is relatively accurate at 98.92% and 94.92% precise. The proposed method is validated using a real-time experimental setup with a FLIR ONE PRO LT iOS Pro-Grade android-fitted infrared thermography camera to capture the experimental setup images in running condition. The three different conditions of the rotating machine are identified as healthy state at 25&#xa0;°C, defective state at 50&#xa0;°C, and defect start to arise state at 45.5&#xa0;°C. These three conditions in the experimental setup are further automated for easy identification of machine state and shown with the help of blinking of three different LED color lights, red color LED represents the defective condition, blue color LED represents the defect starting to arise condition and white color LED represents the healthy condition. The proposed method offers robust accuracy and reliability by aligning it to metrological principles, making it suitable for advanced industry 4.0 setups and industrial applications.</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

The Temperature-Assisted Defect Detection in Rotating Machinery by Using Infrared Thermography Integrated Modified Artificial Neural Network-Based Image Processing: A Step Towards Futuristic Metrology

  • 期刊论文

  • 发表日期：

    2025年3月6日

查看全文
  • Ekta Yadav,
  • V. K. Chawla

摘要

This work presents an advanced metrology method for defect detection in rotating machine components using infrared thermography integrated modified artificial neural network (MANN) based image processing in a real-time manner. The proposed method utilizes the precise, accurate metrological features of infrared thermography, which are highly useful in accurate defect detection. Infrared thermography images are pre-processed by converting them to greyscale images, and then a median filter is used to remove the noise from the images. Thereafter, the proposed MANN method is used via MATLAB program for fault detection and finding the accurate location of the fault. The quantitative analysis is performed on synthetic star images using performance metrics like accuracy, Jaccard similarity index, Dice similarity index, Sensitivity, and Precision. The resulting yield of the proposed method is measured and compared with the yield of various benchmark edge-based segmentation methods from the literature, and it is found that the proposed MANN method is relatively accurate at 98.92% and 94.92% precise. The proposed method is validated using a real-time experimental setup with a FLIR ONE PRO LT iOS Pro-Grade android-fitted infrared thermography camera to capture the experimental setup images in running condition. The three different conditions of the rotating machine are identified as healthy state at 25 °C, defective state at 50 °C, and defect start to arise state at 45.5 °C. These three conditions in the experimental setup are further automated for easy identification of machine state and shown with the help of blinking of three different LED color lights, red color LED represents the defective condition, blue color LED represents the defect starting to arise condition and white color LED represents the healthy condition. The proposed method offers robust accuracy and reliability by aligning it to metrological principles, making it suitable for advanced industry 4.0 setups and industrial applications.