Design optimization using support vector regression



Polynomial regression (PR) and kriging are standard meta-model techniques used for approximate optimization (AO). Support vector regression (SVR) is a new meta-model technique with higher accuracy and a lower standard deviation than existing techniques. In this paper, we propose a sequential approximate optimization (SAO) method using SVR. Inherited latin hypercube design (ILHD) is used as the design of experiment (DOE), and the trust region algorithm is used as the model management technique, both adopted to increase efficiency in problem solving. We demonstrate the superior accuracy and efficiency of the proposed method by solving three mathematical problems and two engineering design problems. We also compare the proposed method with other meta-models such as kriging, radial basis function (RBF), and polynomial regression.


Support vector regression Trust region algorithm Inherited latin-hypercube design Sequential approximate optimization 


  1. [1]
    T. W. Simpson, J. Peplinkski, P. N. Koch and J. K. Allen, Metamodels for computer based engineering design: survey and recommendations, Engineering with Computers 17 (2) (2001) 129–150.MATHCrossRefGoogle Scholar
  2. [2]
    Alex J. Smola and Bernhard Scholkopf, A tutorial on support vector regression, Statistics and computing 14 (3) (2004) 199–222.CrossRefMathSciNetGoogle Scholar
  3. [3]
    Stella M. Clarke, an H. Griebsch, Timothy W. Simpson, Analysis of Support Vector Regression For Approximation of Complex Engineering Analyses, Journal of mechanical design 127 (6) (2005) 1077–1087.CrossRefGoogle Scholar
  4. [4]
    R. Fletcher, An Algorithm for Solving Linearly Constrained Optimization Problems, Mathematical Programming 2 (1) (1972) 133–165.MATHCrossRefMathSciNetGoogle Scholar
  5. [5]
    M. J. D. Powell, Convergence Properties of a Class of Minimization Algorithms, Nonlinear Programming 2, Academic Press, New York, USA, (1975) 1–27.Google Scholar
  6. [6]
    G. G. Wang, Adaptive Response Surface Method Using Inherited Latin Hypercube Design Points, Transactions of the ASME 125 (2) (2003) 210–220.CrossRefGoogle Scholar
  7. [7]
    J. S. Park, Optimal Latin-hypercube Designs for Computer Experiments, Journal of Statistical Planning and Inference 39 (1) (1994) 95–111.MATHCrossRefMathSciNetGoogle Scholar
  8. [8]
    M. H. McKay, R. J. Beckman and W. J. Conover, A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output form a Computer Code, Technometrics 21 (2) (1979) 239–245.MATHCrossRefMathSciNetGoogle Scholar
  9. [9]
    M. C. Shewry, Maximum Entropy Sampling, Journal of Applied Statistics 14 (1987) 165–170.CrossRefGoogle Scholar
  10. [10]
    J. Sacks, W. J. Welch, T. J. Mitchell and H. P. Wynn, Design and Analysis of Computer Experiments, Statistical Science 4 (4) (1989) 409–435.MATHCrossRefMathSciNetGoogle Scholar
  11. [11]
    M. E. Johnson, L. M. Moore and D. Ylvisaker, Minimax and Maximin Distance Design, Journal of Statistical Planning and Inference 26 (2) (1990) 131–148.CrossRefMathSciNetGoogle Scholar
  12. [12]
    K. T. Fang, C. X. MA and P. Winker, Centered L2-Discrepancy of Random Sampling and Latin Hypercube Design, and Construction of Uniform Designs, Mathematics of Computation 71 (237) (2000) 275–296.CrossRefMathSciNetGoogle Scholar
  13. [13]
    G. N. Vanderplaats, Dot Users Manual Version 5.0, VR&D, Colorado, USA, (1999) 12–111.Google Scholar
  14. [14]
    Youn-Hyun Kim, Kyung-Jin Hong and Dong-Hoon Choi, Optimal design of switched reluctance motor using two-dimensional finite element method, Journal of Applied Physics 91 (10) (2002) 6967–6969.CrossRefGoogle Scholar

Copyright information

© Korean Society of Mechanical Engineers 2008

Authors and Affiliations

  1. 1.The Center of Innovative Design Optimization Technology (iDOT)Hanyang UniversitySungdong-Gu, SeoulKorea
  2. 2.The Center of innovative Design Optimization Technology (iDOT)Hanyang UniversitySungdong-Gu, SeoulKorea

Personalised recommendations