Abstract

With the advancement of sensor technologies, data analysis, and computational resources, data-driven models have become important in many different scientific fields. However, for complex problems without having sufficient number of data, the accuracy of the data-driven approaches deteriorates. As an alternative, theory-guided data science which is a combination of physics-driven and data-driven models is a promising approach. In this chapter, a peridynamic machine learning approach is presented based on linear regression. To demonstrate the capability of the coupled peridynamic machine learning approach, four different numerical examples are considered including one-dimensional bar subjected to axial loading, vibration of a one-dimensional bar, two-dimensional plate subjected to tension loading and two-dimensional plate with a pre-existing crack subjected to tension loading.
Original languageEnglish
Title of host publicationPeridynamic Modeling, Numerical Techniques, and Applications
EditorsErkan Oterkus, Selda Oterkus, Erdogan Madenci
Place of PublicationAmsterdam, Netherlands
Chapter20
Pages419-435
Number of pages17
DOIs
Publication statusPublished - 30 Apr 2021

Publication series

NameA volume in Elsevier Series in Mechanics of Advanced Materials

Keywords

  • artificial intelligence
  • linear regression
  • machine learning
  • nonlocal
  • peridynamics

Fingerprint Dive into the research topics of 'Application of artificial intelligence and machine learning in peridynamics'. Together they form a unique fingerprint.

Cite this