Modeling of layer-wise additive manufacturing for part quality prediction

Jianjing Zhang; Peng Wang; Robert X. Gao

doi:10.1016/j.promfg.2018.10.165

Modeling of layer-wise additive manufacturing for part quality prediction

Jianjing Zhang, Peng Wang, Robert X. Gao

Research output: Contribution to journal › Conference article › peer-review

19 Scopus citations

Abstract

Widespread adoption of Additive Manufacturing (AM) has remained a challenge due to part quality inconsistency. Using Fused Deposition Modeling as a representative AM process, this paper presents a deep learning technique termed Long-Short Term Memory for quantification of the nonlinear relationship between the printing process and part tensile strength. The presented modeling method takes into account the layer thermal history associated with the layer-wise printing process as well as the IR sensing data acquired online. Evaluation using Polylactide as the printing material has shown a 46% reduction in prediction error of part tensile strength achieved with the developed modeling method.

Original language	English
Pages (from-to)	155-162
Number of pages	8
Journal	Procedia Manufacturing
Volume	16
DOIs	https://doi.org/10.1016/j.promfg.2018.10.165
State	Published - 2018
Event	7th International Conference on Through-life Engineering Services, TESconf 2018 - Cranfield, United Kingdom Duration: Nov 6 2018 → Nov 7 2018

Bibliographical note

Publisher Copyright:
© 2018 The Authors. Published by Elsevier B.V.

Keywords

Additive Manufacturing
Deep Learning
Predictive Analytics
Process Modeling

ASJC Scopus subject areas

Artificial Intelligence
Industrial and Manufacturing Engineering

Access to Document

10.1016/j.promfg.2018.10.165

Cite this

@article{6c92a07a405e4cf898bbdb5a839fe34e,

title = "Modeling of layer-wise additive manufacturing for part quality prediction",

abstract = "Widespread adoption of Additive Manufacturing (AM) has remained a challenge due to part quality inconsistency. Using Fused Deposition Modeling as a representative AM process, this paper presents a deep learning technique termed Long-Short Term Memory for quantification of the nonlinear relationship between the printing process and part tensile strength. The presented modeling method takes into account the layer thermal history associated with the layer-wise printing process as well as the IR sensing data acquired online. Evaluation using Polylactide as the printing material has shown a 46% reduction in prediction error of part tensile strength achieved with the developed modeling method.",

keywords = "Additive Manufacturing, Deep Learning, Predictive Analytics, Process Modeling",

author = "Jianjing Zhang and Peng Wang and Gao, {Robert X.}",

note = "Publisher Copyright: {\textcopyright} 2018 The Authors. Published by Elsevier B.V.; 7th International Conference on Through-life Engineering Services, TESconf 2018 ; Conference date: 06-11-2018 Through 07-11-2018",

year = "2018",

doi = "10.1016/j.promfg.2018.10.165",

language = "English",

volume = "16",

pages = "155--162",

}

TY - JOUR

T1 - Modeling of layer-wise additive manufacturing for part quality prediction

AU - Zhang, Jianjing

AU - Wang, Peng

AU - Gao, Robert X.

PY - 2018

Y1 - 2018

N2 - Widespread adoption of Additive Manufacturing (AM) has remained a challenge due to part quality inconsistency. Using Fused Deposition Modeling as a representative AM process, this paper presents a deep learning technique termed Long-Short Term Memory for quantification of the nonlinear relationship between the printing process and part tensile strength. The presented modeling method takes into account the layer thermal history associated with the layer-wise printing process as well as the IR sensing data acquired online. Evaluation using Polylactide as the printing material has shown a 46% reduction in prediction error of part tensile strength achieved with the developed modeling method.

AB - Widespread adoption of Additive Manufacturing (AM) has remained a challenge due to part quality inconsistency. Using Fused Deposition Modeling as a representative AM process, this paper presents a deep learning technique termed Long-Short Term Memory for quantification of the nonlinear relationship between the printing process and part tensile strength. The presented modeling method takes into account the layer thermal history associated with the layer-wise printing process as well as the IR sensing data acquired online. Evaluation using Polylactide as the printing material has shown a 46% reduction in prediction error of part tensile strength achieved with the developed modeling method.

KW - Additive Manufacturing

KW - Deep Learning

KW - Predictive Analytics

KW - Process Modeling

UR - http://www.scopus.com/inward/record.url?scp=85060869883&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060869883&partnerID=8YFLogxK

U2 - 10.1016/j.promfg.2018.10.165

DO - 10.1016/j.promfg.2018.10.165

M3 - Conference article

AN - SCOPUS:85060869883

VL - 16

SP - 155

EP - 162

JO - Procedia Manufacturing

JF - Procedia Manufacturing

T2 - 7th International Conference on Through-life Engineering Services, TESconf 2018

Y2 - 6 November 2018 through 7 November 2018

ER -

Modeling of layer-wise additive manufacturing for part quality prediction

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this