OPTIMIZATION OF ELECTROSPINNING PARAMETERS USING AN ARTIFICIAL NEURAL NETWORK (ANN) MODEL FOR ENHANCED NANOFIBER PRODUCTION

Abstract

Electrospinning is a simple and cost-effective technique for creating nanofibers with diverse applications.Optimizing electrospinning parameters is crucial for producing nanofibers with desirable attributes, such as uniform diameter and bead-free morphology.Conventional trial-and-error strategies are frequently protracted and may not necessarily result in optimal outcomes. This investigation delineates the formulation of an artificial neural network (ANN) model specifically designed to systematically optimize electrospinning parameters. Crucial input variables, such as applied voltage, feed rate, and polymer concentration, were utilized to train the ANN model, which was constructed with multiple hidden layers to effectively encapsulate the intricate relationships between input parameters and the resultant nanofiber properties. In this research, an ANN was devised with a 4-3-1 architecture that was trained on a dataset extrapolated from experimental data documented in prior literature and employed the Levenberg-Marquardt algorithm to ascertain robust performance. Upon validation, the model proficiently predicted optimal parameters conducive to the production of smooth, bead-free nanofibers. The model achieved a root mean square error (RMSE) of 7.77%, which is lower than previous models for predicting electrospun Kefiran nanofiber diameter.The results indicate that the ANN-based methodology substantially augments the efficiency and precision of electrospinning parameter optimization, thereby providing a significant resource for researchers and engineers engaged in the domain of nanomaterials. Future investigations could delve into the application of this model to various polymer systems and further refine the ANN architecture to accommodate more intricate electrospinning configurations.