<?xml version="1.0" encoding="utf-8"?>
<export-example>
  <doc>
    <id>5538</id>
    <completedYear/>
    <publishedYear>2022</publishedYear>
    <thesisYearAccepted/>
    <language>eng</language>
    <pageFirst/>
    <pageLast/>
    <pageNumber/>
    <edition/>
    <issue/>
    <volume>215</volume>
    <type>article</type>
    <publisherName>Elsevier</publisherName>
    <publisherPlace/>
    <creatingCorporation/>
    <contributingCorporation/>
    <belongsToBibliography>0</belongsToBibliography>
    <completedDate>--</completedDate>
    <publishedDate>--</publishedDate>
    <thesisDateAccepted>--</thesisDateAccepted>
    <title language="eng">Data-driven models for structure-property prediction in additively manufactured steels</title>
    <abstract language="eng">Data-driven models are developed to predict the mechanical properties of polycrystalline materials. The case study is the prediction of the yield strength of a 3D-printed 316L steel from morphological and crystallographic features. Three different artificial intelligence models including feed-forward (FNN), convolution (CNN), and graph (GNN) neural networks are employed to train the data-driven models and are compared in terms of performance and computational requirements. The dataset required for training is generated by performing crystal plasticity finite element simulations. The FNN model has the smallest input size and takes in some statistical parameters describing the material microstructure, but its accuracy is relatively low. The CNN approach inputs voxel-based realizations of the microstructure and is able to give accurate estimations; however, its training process is time-consuming and computationally expensive. In the GNN approach, the polycrystalline material is represented by a graph whose nodes and lines represent the grains and adjacency between grains. It is observed that GNN yields a better performance compared to the other two approaches and has the capability of handling complex tasks.</abstract>
    <parentTitle language="eng">Computational Materials Science</parentTitle>
    <identifier type="doi">10.1016/j.commatsci.2022.111782</identifier>
    <enrichment key="opus.import.date">2022-10-29T17:49:03+00:00</enrichment>
    <enrichment key="opus.source">sword</enrichment>
    <enrichment key="opus.import.user">importuser</enrichment>
    <enrichment key="BegutachtungStatus">peer-reviewed</enrichment>
    <licence>Keine Lizenz - Es gilt das deutsche Urheberrecht: § 53 UrhG</licence>
    <author>Ebrahim Sadeghpour</author>
    <author>Aida Nonn</author>
    <subject>
      <language>deu</language>
      <type>uncontrolled</type>
      <value>Data-driven model</value>
    </subject>
    <subject>
      <language>deu</language>
      <type>uncontrolled</type>
      <value>Neural networks</value>
    </subject>
    <subject>
      <language>deu</language>
      <type>uncontrolled</type>
      <value>Polycrystalline materials</value>
    </subject>
    <subject>
      <language>deu</language>
      <type>uncontrolled</type>
      <value>Structure-property relationships</value>
    </subject>
    <collection role="institutes" number="FAKMB">Fakultät Maschinenbau</collection>
    <collection role="institutes" number="TCNeustadt">Technologie-Campus Neustadt an der Donau</collection>
    <collection role="persons" number="nonncmm">Nonn, Aida (Prof. Dr.) - Computational Mechanics and Materials Lab</collection>
    <collection role="persons" number="tcnehno">Technologie-Campus Neustadt a. d. Donau (Prof. Ehrlich+ Prof. Nonn)</collection>
    <collection role="othforschungsschwerpunkt" number="16316">Produktion und Systeme</collection>
    <collection role="institutes" number="">Computational Mechanics and Materials Lab (CMM)</collection>
  </doc>
</export-example>
