Multifidelity conditional value-at-risk estimation by dimensionally decomposed generalized polynomial chaos-Kriging

被引:3
作者
Lee, Dongjin [1 ]
Kramer, Boris [1 ]
机构
[1] Univ Calif San Diego, Dept Mech & Aerosp Engn, San Diego, CA 92161 USA
关键词
Risk measures; Conditional Value-at-Risk; Dimensionally decomposed generalized; polynomial chaos expansion; Kriging; Importance sampling; Multifidelity importance sampling; UNCERTAINTY QUANTIFICATION; OPTIMIZATION;
D O I
10.1016/j.ress.2023.109208
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
We propose novel methods for Conditional Value-at-Risk (CVaR) estimation for nonlinear systems under high-dimensional dependent random inputs. We develop a novel DD-GPCE-Kriging surrogate that merges dimensionally decomposed generalized polynomial chaos expansion and Kriging to accurately approximate nonlinear and nonsmooth random outputs. We use DD-GPCE-Kriging (1) for Monte Carlo simulation (MCS) and (2) within multifidelity importance sampling (MFIS). The MCS-based method samples from DD-GPCE-Kriging, which is efficient and accurate for high-dimensional dependent random inputs, yet introduces bias. Thus, we propose an MFIS-based method where DD-GPCE-Kriging determines the biasing density, from which we draw a few high-fidelity samples to provide an unbiased CVaR estimate. To accelerate the biasing density construction, we compute DD-GPCE-Kriging using a cheap-to-evaluate low-fidelity model. Numerical results for mathematical functions show that the MFIS-based method is more accurate than the MCS-based method when the output is nonsmooth. The scalability of the proposed methods and their applicability to complex engineering problems are demonstrated on a two-dimensional composite laminate with 28 (partly dependent) random inputs and a three-dimensional composite T-joint with 20 (partly dependent) random inputs. In the former, the proposed MFIS-based method achieves 104x speedup compared to standard MCS using the high-fidelity model, while accurately estimating CVaR with 1.15% error.
引用
收藏
页数:18
相关论文
共 49 条
  • [1] [Anonymous], 2021, MATLAB VERS 9 10 0 1
  • [2] Cross Validation and Maximum Likelihood estimations of hyper-parameters of Gaussian processes with model misspecification
    Bachoc, Francois
    [J]. COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2013, 66 : 55 - 69
  • [3] Volatility Uncertainty Quantification in a Stochastic Control Problem Applied to Energy
    Bernal, Francisco
    Gobet, Emmanuel
    Printems, Jacques
    [J]. METHODOLOGY AND COMPUTING IN APPLIED PROBABILITY, 2020, 22 (01) : 135 - 159
  • [4] Composite system adequacy assessment using sequential Monte Carlo simulation with variance reduction techniques
    Billinton, R
    Jonnavithula, A
    [J]. IEE PROCEEDINGS-GENERATION TRANSMISSION AND DISTRIBUTION, 1997, 144 (01) : 1 - 6
  • [5] Chaudhuri A., 2020, AIAA SCIT 2020 FOR, P2129, DOI DOI 10.2514/6.2020-2129
  • [6] Certifiable Risk-Based Engineering Design Optimization
    Chaudhuri, Anirban
    Kramer, Boris
    Norton, Matthew
    Royset, Johannes O.
    Willcox, Karen
    [J]. AIAA JOURNAL, 2022, 60 (02) : 551 - 565
  • [7] Chen Xi., 2012, Simulation Conference (WSC), Proceedings of the 2012 Winter, P1
  • [8] Durrande N., 2012, Annales de la Facult des sciences de Toulouse: Mathmatiques Ser. 6, V21, P481, DOI [DOI 10.5802/AFST.1342, 10.5802/afst.1342]
  • [9] Support vector machines based on convex risk functions and general norms
    Gotoh, Jun-ya
    Uryasev, Stan
    [J]. ANNALS OF OPERATIONS RESEARCH, 2017, 249 (1-2) : 301 - 328
  • [10] FAILURE CRITERIA FOR UNIDIRECTIONAL FIBER COMPOSITES
    HASHIN, Z
    [J]. JOURNAL OF APPLIED MECHANICS-TRANSACTIONS OF THE ASME, 1980, 47 (02): : 329 - 334