Abstract
The current research work is the demonstration of the development of a monitoring methodology for the identification of defective weld using surface-level features. A novel grey level-based indicator has been proposed in this research work that has the potential to identify defective welds. The proposed methodology has been validated with a well-established 2D wavelet transform framework to prove the capacity of the proposed methodology. In comparison to the wavelet transform the proposed methodology has been found to be more successful and computationally efficient with less a priori knowledge and limited pre-processing steps making it more effective. In this framework extracted image features along with the process parameters are used as the input to the support vector machine-based regression model for the prediction of the ultimate tensile strength of the welds. It is observed that the model yields an average absolute percentage error of 2.83% compared to the model developed only with process parameters that yield an error of 5.09%. The improvement in the prediction capability of the model leads to the impression that the proposed methodology is effective. The methodology can be used in human free automation system for monitoring of manual metal arc welding process.
Acknowledgements
The authors would like to sincerely express gratitude towards the Department of Mechanical Engineering, National Institute of Technology Silchar, Silchar 788010, Assam, India for providing the facilities to carry out the research work and DST-FIST for sponsoring the instruments at the department. The author also acknowledges Condition Monitoring Laboratory and Advanced Welding Laboratory supported by DST-SERB research project SRG/2020/000293, Department of Mechanical Engineering at National Institute of Technology Silchar for providing the facilities to carry out the research work.
Disclosure statement
The authors declare that they have no conflict of interest.