Experimental Analyses and Predictive Modelling of Ultrasonic Welding Parameters for Enhancing Smart Textile Fabrication

• Published in: Sensors (Basel, Switzerland) Switzerland), 2024-02, Vol.24 (5), p.1488
• Main Authors: Baraya, Mohamed, El-Asfoury, Mohamed S, Fadel, Omnia O, Abass, Ahmed
• Format: Article
• Language: English
• Subjects: Additive manufacturing; Analysis; Artificial intelligence; artificial neural network (ANN); Back propagation; Boundary conditions; Copper; Copper products; copper wire joining; Design of experiments; Electric properties; Flexibility; full factorial experimental design (FFD); Geometry; Heat; Machine learning; Neural networks; Process controls; Screen printing; smart textiles; Software; Textiles; ultrasonic welding; Welding; Wire
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s24051488

This study aims to illustrate the design, fabrication, and optimisation of an ultrasonic welding (UW) machine to join copper wires with non-woven PVC textiles as smart textiles. The study explicitly evaluates UW parameters' impact on heat generation, joint strength, and electrical properties, with a comprehensive understanding of the process dynamics and developing a predictive model applicable to smart textiles. The methodological approach involved designing and manufacturing an ultrasonic piezoelectric transducer using ABAQUS finite element analyses (FEA) software and constructing a UW machine for the current purpose. The full factorial design (FFD) approach was employed in experiments to systematically assess the influence of welding time, welding pressure, and copper wire diameter on the produced joints. Experimental data were meticulously collected, and a backpropagation neural network (BPNN) model was constructed based on the analysis of these results. The results of the experimental investigation provided valuable insights into the UW process, elucidating the intricate relationship between welding parameters and heat generation, joint strength, and post-welding electrical properties of the copper wires. This dataset served as the basis for developing a neural network model, showcasing a high level of accuracy in predicting welding outcomes compared to the FFD model. The neural network model provides a valuable tool for controlling and optimising the UW process in the realm of smart textile production.