Genetic programming-assisted multi-scale optimization for multi-objective dynamic performance of laminated composites: the advantage of more elementary-level analyses
Genetic programming-assisted multi-scale optimization for multi-objective dynamic performance of laminated composites: the advantage of more elementary-level analyses
High-fidelity multi-scale design optimization of many real-life applications in structural engineering still remains largely intractable due to the computationally intensive nature of numerical solvers like finite element method. Thus, in this paper, an alternate route of metamodel-based design optimization methodology is proposed in multi-scale framework based on a symbolic regression implemented using genetic programming (GP) coupled with d-optimal design. This approach drastically cuts the computational costs by replacing the finite element module with appropriately constructed robust and efficient metamodels. Resulting models are compact, have good interpretability and assume a free-form expression capable of capturing the non-linearly, complexity and vastness of the design space. Two robust nature-inspired optimization algorithms, viz. multi-objective genetic algorithm and multi-objective particle swarm optimization, are used to generate Pareto optimal solutions for several test problems with varying complexity. TOPSIS, a multi-criteria decision-making approach, is then applied to choose the best alternative among the Pareto optimal sets. Finally, the applicability of GP in efficiently tackling multi-scale optimization problems of composites is investigated, where a real-life scenario is explored by varying fractions of pertinent engineering materials to bring about property changes in the final composite structure across two different scales. The study reveals that a microscale optimization leads to better optimized solutions, demonstrating the advantage of carrying out a multi-scale optimization without any additional computational burden.
d-Optimal design, Genetic programming, Machine learning-based optimization, Multi-scale optimization, Robust composite structures, Symbolic regression
7969-7993
Kalita, Kanak
b3982788-2ede-4b5a-a606-85d8de81faf2
Mukhopadhyay, Tanmoy
2ae18ab0-7477-40ac-ae22-76face7be475
Dey, Partha
04acb302-d11a-4588-83f7-397313da3613
Haldar, Salil
b01b6cd1-eccb-4cc7-8090-d9b88e28d3d5
1 June 2020
Kalita, Kanak
b3982788-2ede-4b5a-a606-85d8de81faf2
Mukhopadhyay, Tanmoy
2ae18ab0-7477-40ac-ae22-76face7be475
Dey, Partha
04acb302-d11a-4588-83f7-397313da3613
Haldar, Salil
b01b6cd1-eccb-4cc7-8090-d9b88e28d3d5
Kalita, Kanak, Mukhopadhyay, Tanmoy, Dey, Partha and Haldar, Salil
(2020)
Genetic programming-assisted multi-scale optimization for multi-objective dynamic performance of laminated composites: the advantage of more elementary-level analyses.
Neural Computing and Applications, 32 (12), .
(doi:10.1007/s00521-019-04280-z).
Abstract
High-fidelity multi-scale design optimization of many real-life applications in structural engineering still remains largely intractable due to the computationally intensive nature of numerical solvers like finite element method. Thus, in this paper, an alternate route of metamodel-based design optimization methodology is proposed in multi-scale framework based on a symbolic regression implemented using genetic programming (GP) coupled with d-optimal design. This approach drastically cuts the computational costs by replacing the finite element module with appropriately constructed robust and efficient metamodels. Resulting models are compact, have good interpretability and assume a free-form expression capable of capturing the non-linearly, complexity and vastness of the design space. Two robust nature-inspired optimization algorithms, viz. multi-objective genetic algorithm and multi-objective particle swarm optimization, are used to generate Pareto optimal solutions for several test problems with varying complexity. TOPSIS, a multi-criteria decision-making approach, is then applied to choose the best alternative among the Pareto optimal sets. Finally, the applicability of GP in efficiently tackling multi-scale optimization problems of composites is investigated, where a real-life scenario is explored by varying fractions of pertinent engineering materials to bring about property changes in the final composite structure across two different scales. The study reveals that a microscale optimization leads to better optimized solutions, demonstrating the advantage of carrying out a multi-scale optimization without any additional computational burden.
This record has no associated files available for download.
More information
Published date: 1 June 2020
Additional Information:
Funding Information:
KK acknowledges the financial support from MHRD, India, through the award of Ph.D. Scholarship during the period of this research work.
Publisher Copyright:
© 2019, Springer-Verlag London Ltd., part of Springer Nature.
Keywords:
d-Optimal design, Genetic programming, Machine learning-based optimization, Multi-scale optimization, Robust composite structures, Symbolic regression
Identifiers
Local EPrints ID: 483574
URI: http://eprints.soton.ac.uk/id/eprint/483574
ISSN: 0941-0643
PURE UUID: 82c28ee5-31a5-4d83-a477-2123c432ccff
Catalogue record
Date deposited: 01 Nov 2023 18:02
Last modified: 18 Mar 2024 04:10
Export record
Altmetrics
Contributors
Author:
Kanak Kalita
Author:
Tanmoy Mukhopadhyay
Author:
Partha Dey
Author:
Salil Haldar
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics