Radial basis functions as surrogate models with a priori bias in comparison with a posteriori bias

被引:43
作者
Amouzgar, Kaveh [1 ,3 ]
Stromberg, Niclas [2 ]
机构
[1] Jonkoping Univ, Sch Engn, Prod Dev Dept, S-55111 Jonkoping, Sweden
[2] Univ Orebro, Sch Sci & Technol, Dept Mech Engn, S-70182 Orebro, Sweden
[3] Univ Skovde, Sch Engn Sci, POB 408, S-54128 Skovde, Sweden
关键词
Metamodeling; Radial basis function; Design optimization; Design of experiment; METAMODELING TECHNIQUES; ENGINEERING DESIGN;
D O I
10.1007/s00158-016-1569-0
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
In order to obtain a robust performance, the established approach when using radial basis function networks (RBF) as metamodels is to add a posteriori bias which is defined by extra orthogonality constraints. We mean that this is not needed, instead the bias can simply be set a priori by using the normal equation, i.e. the bias becomes the corresponding regression model. In this paper we demonstrate that the performance of our suggested approach with a priori bias is in general as good as, or even for many test examples better than, the performance of RBF with a posteriori bias. Using our approach, it is clear that the global response is modelled with the bias and that the details are captured with radial basis functions. The accuracy of the two approaches are investigated by using multiple test functions with different degrees of dimensionality. Furthermore, several modeling criteria, such as the type of radial basis functions used in the RBFs, dimension of the test functions, sampling techniques and size of samples, are considered to study their affect on the performance of the approaches. The power of RBF with a priori bias for surrogate based design optimization is also demonstrated by solving an established engineering benchmark of a welded beam and another benchmark for different sampling sets generated by successive screening, random, Latin hypercube and Hammersley sampling, respectively. The results obtained by evaluation of the performance metrics, the modeling criteria and the presented optimal solutions, demonstrate promising potentials of our RBF with a priori bias, in addition to the simplicity and straight-forward use of the approach.
引用
收藏
页码:1453 / 1469
页数:17
相关论文
共 27 条
[1]  
Amouzgar K, 2014, ASME 2014 INT DES EN
[2]  
Amouzgar Kaveh, 2013, P ASME 2013 INT DE B, V3B, DOI [10.1115/DETC2013-12809, DOI 10.1115/DETC2013-12809]
[3]  
[Anonymous], 2001, Neural Networks: A Comprehensive Foundation
[4]   A comparative study of the scalability of alternative metamodelling techniques [J].
Backlund, Peter B. ;
Shahan, David W. ;
Seepersad, Carolyn Conner .
ENGINEERING OPTIMIZATION, 2012, 44 (07) :767-786
[5]   ON THE EXPERIMENTAL ATTAINMENT OF OPTIMUM CONDITIONS [J].
BOX, GEP ;
WILSON, KB .
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 1951, 13 (01) :1-45
[7]   A comparative study of metamodeling methods for multiobjective crashworthiness optimization [J].
Fang, H ;
Rais-Rohani, M ;
Liu, Z ;
Horstemeyer, MF .
COMPUTERS & STRUCTURES, 2005, 83 (25-26) :2121-2136
[8]   Recent advances in surrogate-based optimization [J].
Forrester, Alexander I. J. ;
Keane, Andy J. .
PROGRESS IN AEROSPACE SCIENCES, 2009, 45 (1-3) :50-79
[9]   SOLVING STRUCTURAL ENGINEERING DESIGN OPTIMIZATION PROBLEMS USING AN ARTIFICIAL BEE COLONY ALGORITHM [J].
Garg, Harish .
JOURNAL OF INDUSTRIAL AND MANAGEMENT OPTIMIZATION, 2014, 10 (03) :777-794
[10]   ON DESCENT FROM LOCAL MINIMA [J].
GOLDSTEIN AA ;
PRICE, JF .
MATHEMATICS OF COMPUTATION, 1971, 25 (115) :569-574