Scribd
Upload
22 views1 page

Physics Guided Neural Operator For Data Driven Composites Material Process Modelling

This paper presents a physics-guided neural operator designed for predicting high-dimensional temperature histories in the curing process of composite manufacturing. By integrating domain knowledge into a parameterized neural network, the model effectively learns the mapping from cure cycles to temperature histories using limited data, while a novel Fourier residual mapping enhances training efficiency. Experimental results demonstrate the model's accuracy in predicting temperature histories and its potential for optimizing manufacturing processes.

Uploaded by

Isaac Kuma Yeboah
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
22 views1 page

Physics Guided Neural Operator For Data Driven Composites Material Process Modelling

This paper presents a physics-guided neural operator designed for predicting high-dimensional temperature histories in the curing process of composite manufacturing. By integrating domain knowledge into a parameterized neural network, the model effectively learns the mapping from cure cycles to temperature histories using limited data, while a novel Fourier residual mapping enhances training efficiency. Experimental results demonstrate the model's accuracy in predicting temperature histories and its potential for optimizing manufacturing processes.

Uploaded by

Isaac Kuma Yeboah
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Journal of Manufacturing Systems 70 (2023) 217–229

Contents lists available at ScienceDirect

Journal of Manufacturing Systems


journal homepage: www.elsevier.com/locate/jmansys

Physics-guided neural operator for data-driven composites manufacturing


process modelling✩
Gengxiang Chen a,b , Yingguang Li a ,∗, Xu Liu c , Charyar Mehdi-Souzani b , Qinglu Meng a ,
Jing Zhou a , Xiaozhong Hao a
a
College of Mechanical & Electrical Engineering, Nanjing University of Aeronautics and Astronautics, 210016, Nanjing, China
b Université Paris-Saclay, ENS Pars-Saclay, LURPA, Université Sorbonne Paris Nord, 91190, Gif-Sur-Yvette, France
c School of Mechanical and Power Engineering, Nanjing Tech University, 211816, Nanjing, China

ARTICLE INFO ABSTRACT

Keywords: Composite structures have been employed in a variety of applications such as aerospace, automotive, et al. As
Composites manufacturing a vital stage of composites manufacturing, curing involves complex physical and chemical behaviours and
Machine learning plays a significant role in the final mechanical properties and dimensional accuracy of the manufactured
Process modelling
product. Modelling the curing process, especially the real temperature history, is necessary for the optimisation
Smart manufacturing
of cure cycles during composites manufacturing. Since thermochemical analysis using Finite Element (FE)
simulations requires heavy computational loads and data-driven approaches suffer from the complexity of
high-dimensional mapping. This paper proposes a physics-guided neural operator to directly predict the
high-dimensional temperature history from the given cure cycle. By integrating domain knowledge into a
time-resolution independent parameterised neural network, the mapping between cure cycles to temperature
histories can be learned using a limited number of labelled data. Besides, a novel Fourier residual mapping
is designed based on mode decomposition to accelerate the training and significantly boost the performance.
Detailed experiments in several cases show that the proposed model can predict the temperature histories
accurately and also can provide better process optimisation results.

1. Introduction temperature and degree of cure within the composite component which
will further influence the residual stress field and even the performance
Composite materials, such as Carbon Fiber Reinforced Plastic of the final composite structure [3,7]. Therefore, the thermochemical
(CFRP), have been increasingly used in aerospace, automotive and model that predicts fields of temperature and degree of cure is neces-
other modern manufacturing industries because of their high structural sary for designing and optimisation of the cure cycles [8]. Typically,
performance (e.g. strength-to-weight ratio and corrosion resistance) [1, the fields of temperature and degree of cure nowadays are usually
2]. The expansion of demand for composite materials, along with calculated by computational approximation with the Finite Element
increased complexity in geometric shapes and scale, has brought new Method (FEM) software [9]. The cure cycle can be optimised according
challenges to the composite manufacturing industry [3]. To meet the to the simulation results, and an optimised cure cycle usually requires
growing demands for improved product performance and cost-effective massive iterations. But for parts with large sizes and complex structures,
manufacturing, composite manufacturing is undergoing a transfor- the computational efficiency of FEM cannot meet the requirements for
mative shift towards digitisation and intelligence, enabled by recent the iterative optimisation of cure cycles in practical applications [10].
advancements in artificial intelligence and data modelling technologies With the substantial developments of machine learning techniques,
[4,5]. data-driven modelling methods models like neural networks have been
widely used in different manufacturing process modelling scenarios [11,
Manufacturing process modelling aims to construct high-fidelity
12], including composite manufacturing [13]. Recent researchers pro-
predictive representations of the concerned properties of products,
pose to approximate the composite thermochemical model by training
processes or manufacturing systems, for further process optimisation
with the data samples generated from expensive high-fidelity FEM sim-
for decision-making [6]. For composites manufacturing, as shown in
ulations [14,15]. These attempts are meaningful because an accurate
Fig. 1, the cure cycle heavily determines the evolutions of the fields of

✩ The source code of this research is available at github: https://github.com/gengxiangc/ResFNO.


∗ Corresponding author.
E-mail address: [email protected] (Y. Li).

https://doi.org/10.1016/j.jmsy.2023.07.010
Received 21 December 2022; Received in revised form 3 July 2023; Accepted 20 July 2023
Available online 1 August 2023
0278-6125/© 2023 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

You might also like