Application of ANN and RSM techniques for modeling electrospinning process of polycaprolactone

Due to increasing application of nanofibers in many research fields, comprehensive knowledge of the electrospinning process as the most popular method of fiber production is essential. Modeling techniques are valuable tools for managing contributing factors in the electrospinning process, prior to the more expensive experimental techniques. In the present research, effective parameters on the diameter of electrospun polycaprolactone (PCL) nanofibers are analyzed using artificial neural networks (ANN) and response surface methodology (RSM). The assessed parameters include polymer concentration, voltage, and nozzle-to-collector distance. Response surface methodology based on the Box-Behnken design is utilized to develop a mathematical model as well as to determine the optimum condition for production of nanofiber with minimum diameter. In addition, multilayer perceptron neural networks are designed and trained by the sets of input-output patterns using the Levenberg-Marquardt backpropagation algorithm. The high regression coefficient value (R(2)0.97) and low root-mean-square error (RMSE 3.81) of the two models indicate that both models performed well in predicting PCL fiber diameter, although the RSM model slightly outperformed the ANN model in accuracy. The represented models could assist researchers in fabricating electrospun scaffolds with a defined fiber diameter, thus specializing such scaffolds in particular applications.