Welding defect detection using artificial neural network and support vector machine
- Resource Type
- Authors
- Swapnil Gundewar; Prasad Kane; Santosh Behara; Uttam Kumar
- Source
- IOP Conference Series: Materials Science and Engineering. 1259:012029
- Subject
- General Medicine
- Language
- ISSN
- 1757-899X
1757-8981
Welding is an important operation in manufacturing that finds wide applications while joining components. Many destructive and non-destructive techniques are applied to ensure the quality of welded joints. In this paper, an attempt is made to apply the vibration-based technique along with the Artificial Neural Network (ANN) and Support Vector Machine (SVM) to classify the defects of welded joints. Features datasets extracted from the acquired vibration signals during experimentation on the test’s samples fabricated with and without defects are applied to pattern recognition techniques for fault identification. The accuracy of detection of defects using ANN is found to be 90.1% while for SVM it is found to be 92.85% for test datasets. The accuracy of classification obtained for the detection and classification of defects is found to be encouraging demonstrating the suitability of the proposed vibration-based approach to the development of a decision support system for non-destructive testing for defect identification.