Computational Technology Resources - CCP

Keywords: global optimization technique, kriging metamodel, metamodel-based design optimization, statistical global optimization, non-gradient optimization.

Summary

Global optimization techniques have gained much attention in the design of industrial products because they can find not only the best design solution within the design space, but also provide engineers with multiple alternatives. However, considerable the computational burden of global optimization, it is still a challenging problem. In this paper, the statistical global optimization technique using the kriging metamodel is proposed in order to resolve the difficulty. Because the proposed method explores a global optimum by means of the kriging metamodel that can evaluate inexpensively and quickly the responses, the computational cost of global optimization can be remarkably reduced. For mathematical problems with strong nonlinearity, it is shown that the proposed method can search for a global optimum accurately as well as efficiently. In this paper, we propose a useful sampling criterion that can satisfy the feasibility, descent condition of the cost function and space-filling sampling. After clarifying a proposed sampling criterion in detail, the kriging metamodel is explained. Next, a new searching algorithm to efficiently achieve a global optimum is proposed based on the sequential sampling criterion. We named the method the kriging metamodel based statistical global optimization (kriGO). The performance of the developed algorithm for nonlinear mathematical problems are illustrated. Issues such as the ability to solve a problem, convergence properties and efficiency are discussed. As a practical application of the method developed, an engineering design problem of structural optimization for a bogie frame is performed. Through the research, it has been concluded that kriGO can find a global optimum accurately as well as efficiently. Further, the proposed method can search for individual multiple global optima at the same time even though values of objective function are identical to each other.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £145 +P&P)

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Proceedings ISSN 1759-3433 CCP: 93 PROCEEDINGS OF THE TENTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STRUCTURES TECHNOLOGY Edited by: Paper 134 Kriging Metamodel Based Statistical Global Optimization Technique T.H. Lee¹ and S.K. Cho² ¹School of Mechanical Engineering, ²Department of Automotive Engineering, Graduate School Hanyang University, Seoul, Republic of Korea doi:10.4203/ccp.93.134 purchase the full-text of this paper Full Bibliographic Reference for this paper T.H. Lee, S.K. Cho, "Kriging Metamodel Based Statistical Global Optimization Technique", in , (Editors), "Proceedings of the Tenth International Conference on Computational Structures Technology", Civil-Comp Press, Stirlingshire, UK, Paper 134, 2010. doi:10.4203/ccp.93.134 Keywords: global optimization technique, kriging metamodel, metamodel-based design optimization, statistical global optimization, non-gradient optimization. Summary Global optimization techniques have gained much attention in the design of industrial products because they can find not only the best design solution within the design space, but also provide engineers with multiple alternatives. However, considerable the computational burden of global optimization, it is still a challenging problem. In this paper, the statistical global optimization technique using the kriging metamodel is proposed in order to resolve the difficulty. Because the proposed method explores a global optimum by means of the kriging metamodel that can evaluate inexpensively and quickly the responses, the computational cost of global optimization can be remarkably reduced. For mathematical problems with strong nonlinearity, it is shown that the proposed method can search for a global optimum accurately as well as efficiently. In this paper, we propose a useful sampling criterion that can satisfy the feasibility, descent condition of the cost function and space-filling sampling. After clarifying a proposed sampling criterion in detail, the kriging metamodel is explained. Next, a new searching algorithm to efficiently achieve a global optimum is proposed based on the sequential sampling criterion. We named the method the kriging metamodel based statistical global optimization (kriGO). The performance of the developed algorithm for nonlinear mathematical problems are illustrated. Issues such as the ability to solve a problem, convergence properties and efficiency are discussed. As a practical application of the method developed, an engineering design problem of structural optimization for a bogie frame is performed. Through the research, it has been concluded that kriGO can find a global optimum accurately as well as efficiently. Further, the proposed method can search for individual multiple global optima at the same time even though values of objective function are identical to each other. purchase the full-text of this paper (price £20) go to the previous paper go to the next paper return to the table of contents return to the book description purchase this book (price £145 +P&P)
Back to top	©Civil-Comp Limited 2023 - terms & conditions