Int J Performability Eng ›› 2019, Vol. 15 ›› Issue (1): 107-115.

Fitting Methods based on Custom Neural Network for Relaxation Modulus of Viscoelastic Materials

Yun Heab, Haibin Lia*(), and Juan Dua

1. a College of Science,Inner Mongolia University of Technology,Hohhot,010051,China
b College of Water Conservancy and Civil Engineering,Inner Mongolia Agricultural University,Hohhot,010018,China
•  Revised on  ; Accepted on
• Contact: Li Haibin E-mail:lhbnm2003@126.com

Abstract:

The viscoelastic material constitutive relationship is relatively complex in the engineering practice. People often use the Prony series method to fit experimental data. A lower number of terms leads to lower accuracy, but a higher number of terms leads to difficulty of fitting.Thus, a custom neural network model was put forward to replace the traditional algorithm in the Prony series fitting process. Based on each specific form of the Prony series constructed corresponding activation function of the hidden layer in the neural network, the number of neurons in the neural network corresponded to the number of the Prony series. A numerical example showed that the custom neural network can achieve good fitting results. It is convenient and appropriate to select the number of terms and also shows rapid convergence and high accuracy.