AI Summary & Key Takeaways
AI InsightThese results indicate that seam strength can be estimated without physical ultrasonic sewing, enabling faster material screening and reducing experimental iterations during process-window selection and product development.
📌 3 Key Takeaways
Abstract
The demand for protective textiles surged during the COVID-19 pandemic, accelerating the adoption of ultrasonic sewing for nonwoven assemblies. This study develops an artificial neural network (ANN) to predict the ultrasonic seam tensile strength of thermally bonded nonwoven fabrics from fabric, fibre, and process parameters. The model uses 15 inputs spanning fibre composition and properties, calender conditions, fabric basis weight and tensile properties, and anvil-wheel (roller) configuration, and is evaluated on previously unseen test samples. The ANN attains R² ≈ 0.84, MAE = 5.17 N, RMSE = 6.56 N (MSE = 43.00 N²), demonstrating strong standardisation beyond the training data. Consistent with industrial practice and the directional anisotropy of such materials, the study focuses on the machine direction (MD), which governs cutting and seam design in production; cross-direction (CD) testing was not included due to very low loads and poor reproducibility for the investigated material set. These results indicate that seam strength can be estimated without physical ultrasonic sewing, enabling faster material screening and reducing experimental iterations during process-window selection and product development. The approach provides a practical decision aid for the manufacturing of medical and hygiene nonwovens and can be readily integrated with feature-importance analyses to guide parameter prioritisation.