Monitoring of friction stir welding based on vision system coupled with Machine learning algorithm. (October 2019)

• Record Type: Journal Article
• Authors: Sudhagar, S.; Sakthivel, M.; Ganeshkumar, P.
• Is Part Of: Measurement. Volume 144(2019)
• Page Start: 135
• Page End: 143
• Publication Date: 2019-10
• Subjects: Friction stir welding -- Machine learning -- Image processing -- Support Vector Machine; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2019.05.018
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A monitoring technique is proposed for FSW process using surface image. The variation in surface appearance shows good correlation with tensile strength. Features from surface image were extracted using MSER algorithm. Support Vector Machines classifies the good and bad weld with accuracy of 95.8%. Abstract: The increase in utilization of FSW process demands online monitoring system for early detection and control of defects. This research attempts to develop a system for detection and classification of defective welds using weld surface image. Welding joints are produced at different welding condition by varying tool rotational speed, welding speed, tool shoulder diameter and pin diameter. The weld surfaces produced at different welding condition are captured using digital camera and processed to extract features. The features from weld surface image has been extracted using maximally stable extremal region algorithm and which is used as input for classification of weld joint. The Support Vector Machines is used for classification of weld using features from surface image. Support Vector Machines is trained with different kernel functions and found that linear and quadratic kernel function classify defect weld and good weld with accuracy of 95.8%.