Prediction of the fiber diameter of melt electrospinning writing by kriging model

Melt electrospinning writing (MEW) has proven its potential to direct-write complex and multiscaled architectures and structures, which are widely used in the biomedical fields such as 3D printing of porous scaffolds. For the resolution of the microstructure is characterized by the diameter of the melt electrospun fiber, uniform spinning fiber with predictable diameter is of great significance for precise fabrication of the microstructure. In this paper, the Kriging model was introduced to explore the scaling laws of the fiber diameter to several processing parameters (including temperature, tip-to-collector distance, flow rate, and collector speed). The Latin hypercube sampling was adopted for experiment design. The prediction results of the Kriging model were cross-validated with the response surface model to compare the prediction accuracy of the two methods. The root-mean-square error, maximum absolute error, and mean absolute percentage error of the Kriging model (0.49%, 0.73%, and 3.52%, respectively) are all lower than the response surface model (0.81%, 1.25%, and 4.74%, respectively), indicating that the prediction accuracy of the Kriging model is superior to the response surface model. The present paper provides a new idea for the modeling of the complex nonlinear system of MEW. Moreover, the implementation of the Kriging model minimizes the number of experimental trials required from 31 to 27, resulting in the reduction of the fabrication time and materials wastage.

Automated summary

Melt electrospinning writing (MEW) is a powerful technique for direct-writing complex and multiscaled structures, particularly in the biomedical field. This paper introduces the Kriging model to study the relationship between fiber diameter and various processing parameters, and demonstrates that the Kriging model outperforms the response surface model in terms of prediction accuracy. Additionally, implementing the Kriging model reduces fabrication time and material waste.