A physics-guided machine learning model for two-dimensional structures based on ordinary state-based peridynamics

Cong Tien Nguyen, Selda Oterkus, Erkan Oterkus

Research output: Contribution to journalArticlepeer-review

19 Citations (Scopus)
45 Downloads (Pure)

Abstract

This study presents a novel physics-guided machine learning model for two-dimensional structures based on ordinary state-based peridynamics. The linear relationships between the displacements of a material point and the displacements of its neighbours and the applied forces are obtained for the machine learning model by using linear regression. The numerical procedure for coupling the ordinary state-based peridynamic model and the machine learning model is also provided. The accuracy of the coupled model is verified by predicting deformations of a two-dimensional plate with circular cut-out subjected to tension and a two-dimensional representation of three points bending test. To further demonstrate the capabilities of the coupled model, damage predictions for a two-dimensional representation of a three-point bending test, a notched plate with a hole subjected to tension, a square plate with a pre-existing crack subjected to tension, and a plate with a pre-existing crack subjected to sudden loading are presented.
Original languageEnglish
Article number102872
Number of pages31
JournalTheoretical and Applied Fracture Mechanics
Volume112
Early online date18 Jan 2021
DOIs
Publication statusPublished - 30 Apr 2021

Keywords

  • machine learning
  • peridynamics
  • fracture
  • nonlocal
  • linear regression

Fingerprint

Dive into the research topics of 'A physics-guided machine learning model for two-dimensional structures based on ordinary state-based peridynamics'. Together they form a unique fingerprint.

Cite this