東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Improving Robust Design Optimization...

Gupta, Subham,

FindBook

Google Book

Amazon

博客來

Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures // Subham Gupta.
作者:	Gupta, Subham,
面頁冊數:	1 electronic resource (168 pages)
附註:	Source: Dissertations Abstracts International, Volume: 85-12, Section: B.
Contained By:	Dissertations Abstracts International85-12B.
標題:	Aerospace engineering. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30994291
ISBN:	9798382812793

Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures /
Gupta, Subham,

Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures /Subham Gupta. - 1 electronic resource (168 pages)

Source: Dissertations Abstracts International, Volume: 85-12, Section: B.

Robust Design Optimization (RDO) is an approach that addresses the effect of randomness in the design process by considering the uncertainty in the inputs and minimizing the variation in the outputs. A robust design is essential in structural optimization as it ensures that the optimized design is robust to uncertainties in material properties, loads, and other environmental factors. However, performing RDO is a highly complex process, and therefore, new approaches must be developed to reduce computational costs. Machine learning (ML) techniques have significantly impacted structural optimization by providing new approaches to modeling complex relationships between design variables, objectives, and constraints. Among the prevalent methods for ML, Artificial Neural Networks (ANN) have become very popular and are used in performing optimization across various disciplines. Though ANN has been successful in applications with large amounts of training data, applying them to problems with limited data is complicated and needs further research and improvement. Thus, this dissertation investigates the enhancement of RDO performance in structural engineering through advancements in ML techniques, focusing on two primary objectives: 1) Developing new techniques for efficient RDO and 2) Improving ANN training performance while reducing computational costs. Firstly, an efficient RDO approach is proposed to significantly lower the computational expenses typically associated with traditional RDO methods. The proposed strategy involves deterministic optimization of a problem with random variables, followed by uncertainty quantification (UQ) at the obtained optimal solutions. Secondly, a novel optimal-sampling-based, multi-response ANN framework with adaptive modification of the network's hyperparameters is proposed to reduce the number of samples required to build an ANN. The methodology employs the D-optimal design, a sampling method that aims to maximize the determinant of the information matrix. Additionally, the framework introduces a strategy for updating the network, wherein the model's hyperparameters are fine-tuned using a Gaussian process-based Bayesian optimization approach. Finally, an ANN-based RDO framework is implemented for the optimization of a Variable Stiffness Composite (VSC) laminate with a central hole. The framework incorporates two key manufacturing constraints: the minimum radius the automatic fiber placement (AFP) manufacturing head can navigate and manufacturing defects like tow gaps and overlaps.

English

ISBN: 9798382812793Subjects--Topical Terms:

1002622
Aerospace engineering.
Subjects--Index Terms:

Artificial neural networks

Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures /
LDR:04156nmm a22004573i 4500 001 2438530
005 20260727134352.5
006 m o d
007 cr|nu||||||||
008 261011s2024 miu||||||m |||||||eng d
020 $a 9798382812793
035 $a (MiAaPQD)AAI30994291
035 $a AAI30994291
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Gupta, Subham, $e author. $3 3829870
245 1 0 $a Improving Robust Design Optimization Performance Using Machine Learning and Its Application to Composite Structures / $c Subham Gupta.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (168 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-12, Section: B.
500 $a Advisors: Mulani, Sameer B. Committee members: Barkey, Mark E.; Vikas, Vishesh; Goyal, Vijay; Hu, Chongze.
502 $b Ph.D. $c The University of Alabama $d 2024.
520 $a Robust Design Optimization (RDO) is an approach that addresses the effect of randomness in the design process by considering the uncertainty in the inputs and minimizing the variation in the outputs. A robust design is essential in structural optimization as it ensures that the optimized design is robust to uncertainties in material properties, loads, and other environmental factors. However, performing RDO is a highly complex process, and therefore, new approaches must be developed to reduce computational costs. Machine learning (ML) techniques have significantly impacted structural optimization by providing new approaches to modeling complex relationships between design variables, objectives, and constraints. Among the prevalent methods for ML, Artificial Neural Networks (ANN) have become very popular and are used in performing optimization across various disciplines. Though ANN has been successful in applications with large amounts of training data, applying them to problems with limited data is complicated and needs further research and improvement. Thus, this dissertation investigates the enhancement of RDO performance in structural engineering through advancements in ML techniques, focusing on two primary objectives: 1) Developing new techniques for efficient RDO and 2) Improving ANN training performance while reducing computational costs. Firstly, an efficient RDO approach is proposed to significantly lower the computational expenses typically associated with traditional RDO methods. The proposed strategy involves deterministic optimization of a problem with random variables, followed by uncertainty quantification (UQ) at the obtained optimal solutions. Secondly, a novel optimal-sampling-based, multi-response ANN framework with adaptive modification of the network's hyperparameters is proposed to reduce the number of samples required to build an ANN. The methodology employs the D-optimal design, a sampling method that aims to maximize the determinant of the information matrix. Additionally, the framework introduces a strategy for updating the network, wherein the model's hyperparameters are fine-tuned using a Gaussian process-based Bayesian optimization approach. Finally, an ANN-based RDO framework is implemented for the optimization of a Variable Stiffness Composite (VSC) laminate with a central hole. The framework incorporates two key manufacturing constraints: the minimum radius the automatic fiber placement (AFP) manufacturing head can navigate and manufacturing defects like tow gaps and overlaps.
546 $a English
590 $a School code: 0004
650 4 $a Aerospace engineering. $3 1002622
650 4 $a Mechanics. $3 525881
650 4 $a Materials science. $3 543314
653 $a Artificial neural networks
653 $a Machine learning
653 $a Reliability engineering
653 $a Robust design optimization
653 $a Uncertainty quantification
653 $a Variable stiffness composites
690 $a 0538
690 $a 0346
690 $a 0800
690 $a 0794
710 2 $a The University of Alabama. $b Aerospace Engineering. $e degree granting institution. $3 3823887
720 1 $a Mulani, Sameer B. $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 85-12B.
790 $a 0004
791 $a Ph.D.
792 $a 2024
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30994291