Machine learning-based fibre diameter prediction using electrospinning parameters

Electrospinning is a technique used to create fibres, where the electrical force is applied to the needle so that the polymer solution flow is converted to thread fibres and collected in a flat collector. These fibre sizes typically range from a few micrometers to tens of nanometers. Electrospinning is widely used in biomedical applications, including drug delivery systems, tissue engineering, and wound dressings. These electrospun fibres are used in textiles and garments as they offer improved comfort, moisture control, and breathability. Additionally, fibres can be functionalised to give fabrics qualities like antibacterial characteristics, water resistance, and UV protection. Currently, the fibres produced in this technique involve various parameter tuning to get the desired fibre diameter. In this work, we employed machine learning algorithms like support vector machines (SVMs), random forests, XGBoost, gradient boosting regressors, and k nearest neighbour to better predict the fibre diameter using the various electrospinning parameters. Comparing these model algorithms, we found that the Gradient Boosting algorithm gives the best accuracy of 94.8% and minimum error.