14th International Probabilistic Workshop pp 63-80 | Cite as
Moving Least Squares Metamodels—Hyperparameter, Variable Reduction and Model Selection
Abstract
The objective of metamodel applications is to obtain a large amount of system information from a small data set. Areas of application within the Computer-aided engineering are e.g. optimization problems, robust design engineering or sensitivity analysis. This paper deals with the metamodel techniques Least Squares (LS) regression and Moving Least Squares (MLS) as well as with their application in case of multivariate and nonlinear system behavior. In this context, LS regression represents a widely used method, which is limited in application due to the fixed polynomial order and the resulting relationship between existing support points and necessary polynomial coefficients. A more flexible metamodel technique regarding the description of nonlinearities is the MLS approach. In this procedure, the support points are weighted to build a local polynomial. The multivariate MLS-application is implemented by an anisotropic distance measure and a variable reduction. The selection of the most appropriate metamodel is tested for a deterministic model framework of mathematical test functions regarding the polynomial order, variable reduction and metamodel technique.
Keywords
Root Mean Square Error Variable Reduction Weight Little Square Support Point Move Little SquareNotes
Acknowledgments
The investigations presented in this paper were conducted within the German collaboration program COOREFLEX-turbo in the frame of AG Turbo. The work was supported by the Bundesministerium für Wirtschaft und Energie (BMWi) under Grant No. 03ET7041E. The authors gratefully acknowledge AG Turbo and the industrial partners MTU Aero Engines, Rolls-Royce Deutschland and Ansaldo Energia for their support and permission to publish this paper. The responsibility for the content lies solely with its authors.
References
- Arlot S, Celisse A (2010) A survey of cross-validation procedures for model selection. Stat Surv 4:40–79MathSciNetCrossRefMATHGoogle Scholar
- Beschorner A, Voigt M, Vogeler K (2014) Monte_carlo_cross-validation for response surface benchmarkGoogle Scholar
- Cheng Z-Q, Wang Y-Z, Li B, Xu K, Dang G, Jin S-Y(2008) A survey of methods for moving least squares surfaces. In: Volume graphics, pp 9–23Google Scholar
- Cremanns K (2014) Requirements and new approaches of probabilistic optimal design from a practical point of view considering steam turbinesGoogle Scholar
- Cremanns K (2016) Efficient multi-objective optimization of labyrinth seal leakage in steam turbines based on hybrid surrogate models. In: Proceedings of ASME Turbo Expo 2016: turbomachinery technical conference and exposition, Seoul, South KoreaGoogle Scholar
- Cremanns K, Roos D (2016) A new optimized anisotropic moving least squares surrogate model with maximized prognosis. Crete Island, GreeceGoogle Scholar
- Fahrmeir L, Kneib T (2009) Stefan Lang. Springer, RegressionGoogle Scholar
- Forrester AIJ, Sóbester A, Keane AJ (2008) Engineering design via surrogate modelling: a practical guide. Wiley, Chichester, West Sussex, England, Hoboken, NJGoogle Scholar
- Friedman J, Hastie T, Tibshirani R (2001) The elements of statistical learning, vol 1. Springer series in statistics Springer, BerlinMATHGoogle Scholar
- Giunta AA, Swiler LP, Brown SL, Eldred MS, Richards MD, Cyr EC (2006) The surfpack software library for surrogate modeling of sparse irregularly spaced multidimensional data. In: 11th AIAA multidisciplinary analysis and optimization conferenceGoogle Scholar
- Kang S-C, Koh H-M, Choo JF (2010) An efficient response surface method using moving least squares approximation for structural reliability analysis. Probab Eng Mech 25(4):365–371CrossRefGoogle Scholar
- Kiusalaas J (2010) Numerical methods in engineering with Python. Cambridge University Press, New YorkCrossRefMATHGoogle Scholar
- Mehmood T (2012) Kristian hovde liland, lars snipen, and solve sæbø. A review of variable selection methods in partial least squares regression. Chemometr Intell Lab Syst 118:62–69CrossRefGoogle Scholar
- Nealen A (2004) An as-short-as-possible introduction to the least squares, weighted least squares and moving least squares methods for scattered data approximation and interpolation, pp 130–150. http://www.nealen.com/projects
- Nelder JA, Mead R (1965) A simplex method for function minimization. Comput J 7(4):308–313MathSciNetCrossRefMATHGoogle Scholar
- Polynkin A, Toropov V (2010) Recognition of design variable inter-dependencies using cross-validated moving least-squares methodGoogle Scholar
- Rocha H (2009) On the selection of the most adequate radial basis function. Appl Math Model 33(3):1573–1583CrossRefGoogle Scholar
- Ryberg A-B, Domeij Bäckryd R, Nilsson L (2012) Metamodel-based multidisciplinary design optimization for automotive applicationsGoogle Scholar
- Taflanidis AA, Jia G, Kennedy AB, Smith JM (2013) Implementation/optimization of moving least squares response surfaces for approximation of hurricane/storm surge and wave responses. Nat Hazards 66(2):955–983CrossRefGoogle Scholar
- Tian L, Zhenzhou L, Hao W (2013) Moving least squares based sensitivity analysis for models with dependent variables. Appl Math Model 37(8):6097–6109MathSciNetCrossRefMATHGoogle Scholar
- Tu J (2003) Cross-validated multivariate metamodeling methods for physics-based computer simulations. In: Proceedings of the IMAC-XXIGoogle Scholar
- Viana FAC, Gogu C, Haftka RT (2010) Making the most out of surrogate models: tricks of the tradeGoogle Scholar
- Wang GG, Shan S (2006) Review of metamodeling techniques in support of engineering design optimizationGoogle Scholar
- Yomsatieankul W (2010) High-order non-oscillatory schemes using meshfree interpolating moving least squares reconstruction for hyperbolic conservation laws. PhD thesisGoogle Scholar