Related Experiment Video
Updated: Oct 17, 2025
10:10
Three-Dimensional Particle Shape Analysis Using X-ray Computed Tomography: Experimental Procedure and Analysis Algorithms for Metal Powders
Published on: December 4, 2020
1.9K
Microstructural porosity segmentation using machine learning techniques in wire-based direct energy deposition of
Pavan Kumar Nalajam1, Ramesh V1
1School of Electrical Engineering, Vellore Institute of Technology (VIT), Vellore, 632014, India.
Micron (Oxford, England : 1993)
|October 10, 2021
Summary
This study introduces machine learning for detecting porosity in additively manufactured aluminum alloy 6061 components, achieving high accuracy even with limited data. This advancement aids quality control in the aviation industry.
Area of Science:
- Materials Science and Engineering
- Additive Manufacturing
- Artificial Intelligence
Background:
- Additive manufacturing (AM) offers a flexible and economical alternative to traditional methods, transforming industries like aviation.
- Wire Arc Additive Manufacturing (WAAM) enhances efficiency and enables custom components but faces challenges like microstructural porosity.
- Artificial Intelligence (AI) shows potential for porosity detection, yet data scarcity limits its application.
Purpose of the Study:
- To develop and demonstrate machine learning models for effective porosity detection in WAAM aluminum alloy 6061.
- To address the challenge of limited datasets in applying AI for microstructural defect analysis.
- To achieve high accuracy in identifying and classifying pores within additively manufactured materials.
Main Methods:
- Utilized machine learning models for porosity detection in microstructural images of WAAM aluminum alloy 6061.
- Employed Gabor filters to extract pixel-level color and texture features for pore segmentation.
- Optimized model hyperparameters through cross-validation to enhance performance.
Main Results:
- Achieved an average classification accuracy of 98.89% using a random forest model for porosity detection.
- Successfully detected pores larger than 5 μm in microstructural images.
- Demonstrated the effectiveness of the proposed methods compared to existing techniques.
Conclusions:
- Machine learning models are highly effective for porosity detection in WAAM aluminum alloy 6061, even with limited data.
- The developed approach offers a robust solution for quality control in additively manufactured components.
- This research paves the way for wider adoption of AI in analyzing microstructural defects in AM parts.
