Designing and Manufacturing of an Electrospinning Machine and Prediction of Mechanical Responses of Artificial Blood Vessels Using Neural Networks and Finite Element Analysis

Rafiei, Soroush | 2021

375 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53745 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Nourani, Amir; Chizari, Mahmoud
  7. Abstract:
  8. Cardiovascular disease is one of the most leading causes of death in the world and one of the common methods for treating such diseases is replacing damaged blood vessels with artificial ones. Electrospinning, which is a method of producing polymer fibers, is one of the best and common methods of producing synthetic vessels. The first goal in this study was to create a suitable structure for fiber production; therefore, first, an electrospinning device and environmental chamber were designed and built. Then, the effect of alternative and uniform flow on the diameter and structure of the fibers was investigated using 15% wt. Polycaprolactone solution. The results of this study showed that the average of the lowest diameter obtained using a peristaltic pump was 85.57 ± 14.31 nm, while this value was 128.11 ± 8.82 nm using a syringe pump (±95% confidence interval). Also, the fibers produced with alternating flow had breaks, which would cause the use of this flow unsuitable for producing the artificial vessels. In the next step, due to the great importance in the similarity of the compliance of artificial vessels to real vessels, a number of finite element models were used to fully characterize the vessels behavior and the effects of internal diameter, thickness, curvature and Young vessel modulus on the compliance were investigated. Finally, using a trial and error, the best multilayer neural network was obtained to predict the compliance of artificial vessels. The best neural network architecture was found to include a hidden layer and five neurons with the Levenberg-Marquardt training method. This network had a correlation coefficient of 0.996 with a mean squared error of 1.065.

  9. Keywords:
  10. Artificial Neural Network ; Electrospinning ; Finite Element Analysis ; Heart Diseases ; Artificial Vessels ; Cardiovascular System

 Digital Object List