Robust Parameter Design Based on Functional Latent Gaussian Processes and Its Application in Additive Manufacturing
DOI:
https://doi.org/10.54097/twbvbm92Keywords:
Gaussian Process, Robust Parameter Design, Functional Response, Additive Manufacturing.Abstract
For the robust parameter design problem of functional responses, a novel robust parameter optimization model is proposed within the functional latent Gaussian process (LFGP) modeling framework. First, maximum likelihood estimation is employed to obtain the required hyperparameters for fitting the LFGP model. Second, leveraging the characteristics of the LFGP model, we construct an objective function that fully accounts for response variability, thereby establishing an optimization model for functional responses. Finally, we employ a genetic algorithm (GA) for global optimization to obtain optimal input parameter settings. Results from an additive manufacturing case study demonstrate that this method achieves superior prediction accuracy and optimization performance for functional data compared to traditional Gaussian process functional regression and principal component analysis models.
Downloads
References
[1] YAN, Jingdong; MA, Shenghua. Research on evaluation of high-end equipment manufacturing achievements from the perspective of scientific and technological achievements transformation. In: International Conference on Education, Management, and Computer. 2019. p. 185 - 194.
[2] TAGUCHI G. Introduction to quality engineering: designing quality into products and processes [M]. 1986.
[3] CHATURVEDI, Saurabh, et al. Evaluation of the methods for determining accuracy of fit and precision of RPD framework in Digital (3D printed, milled) and conventional RPDs-a systematic review. BMC Oral Health, 2024, 24.1: 1466.
[4] Han, M., Ouyang, L. Robust functional response-based metamodel optimization considering both location and dispersion effects for aeronautical airfoil designs. Struct Multidisc Optim 64, 1545 – 1565 (2021).
[5] Marc C. Kennedy, Anthony O'Hagan, Bayesian Calibration of Computer Models, Journal of the Royal Statistical Society Series B: Statistical Methodology, Volume 63, Issue 3, September 2001, Pages 425 – 464
[6] SHI J, WANG B, MURRAY-SMITH R, et al. Gaussian process functional regression modeling for batch data [J]. Biometrics, 2007, 63 (3): 714 - 23.
[7] LIU F, WEST M. A dynamic modelling strategy for Bayesian computer model emulation [J]. 2009.
[8] BAYARRI M J, BERGER J O, KENNEDY M C, et al. Predicting vehicle crashworthiness: Validation of computer models for functional and hierarchical data [J]. Journal of the American Statistical Association, 2009, 104 (487): 929 - 43.
[9] JOSEPH V R, HUNG Y, SUDJIANTO A. Blind kriging: A new method for developing metamodels [J]. 2008.
[10] ROUGIER J. Efficient emulators for multivariate deterministic functions [J]. Journal of Computational and Graphical Statistics, 2008, 17 (4): 827 - 43.
[11] MELKUMYAN A, RAMOS F. Multi-kernel Gaussian processes; proceedings of the IJCAI Proceedings-International Joint Conference on Artificial Intelligence, F, 2011 [C].
[12] LI B, GENTON M G, SHERMAN M. Testing the covariance structure of multivariate random fields [J]. Biometrika, 2008, 95 (4): 813 - 29.
[13] CHUNG S, KONTAR R. Functional principal component analysis for extrapolating multistream longitudinal data [J]. IEEE Transactions on Reliability, 2020, 70 (4): 1321 - 31.
[14] SUNG C-L, WANG W, PLUMLEE M, et al. Multiresolution functional ANOVA for large-scale, many-input computer experiments [J]. Journal of the American Statistical Association, 2020.
[15] CHEN T, HADINOTO K, YAN W, et al. Efficient meta-modelling of complex process simulations with time–space-dependent outputs [J]. Computers & chemical engineering, 2011, 35 (3): 502 - 9.
[16] DEL CASTILLO E, COLOSIMO B M, ALSHRAIDEH H. Bayesian modeling and optimization of functional responses affected by noise factors [J]. Journal of Quality Technology, 2012, 44 (2): 117 - 35.
[17] KONZEN E, CHENG Y, SHI J Q. Gaussian process for functional data analysis: The GPFDA package for R [J]. arXiv preprint arXiv:210200249, 2021.
[18] MA P, MONDAL A, KONOMI B A, et al. Computer model emulation with high-dimensional functional output in large-scale observing system uncertainty experiments [J]. Technometrics, 2022, 64 (1): 65 - 79.
[19] LU J, ZHAN Z, APLEY D W, et al. Uncertainty propagation of frequency response functions using a multi-output Gaussian Process model [J]. Computers & Structures, 2019, 217: 1 - 17.
[20] FANG X, PAYNABAR K, GEBRAEEL N. Multistream sensor fusion-based prognostics model for systems with single failure modes [J]. Reliability Engineering & System Safety, 2017, 159: 322 - 31.
[21] GUO X, JI C, LIU R, et al. A two-stage approach for frequency response modeling and metamaterial rapid design [J]. Progress In Electromagnetics Research C, 2017, 76: 11 - 22.
[22] LIU Z, LI Y, YUE X, et al. Latent functional Gaussian process incorporating output spatial correlations [J]. IISE Transactions, 2024: 1 - 14.
[23] JU-LONG D. Control problems of grey systems [J]. Systems & control letters, 1982, 1 (5): 288 - 94.
[24] HOLLAND J H. Adaptation in natural and artificial systems: an introductory analysis with applications to biology, control, and artificial intelligence [M]. MIT press, 1992.
[25] CAMPBELL K, MCKAY M D, WILLIAMS B J. Sensitivity analysis when model outputs are functions [J]. Reliability Engineering & System Safety, 2006, 91 (10 - 11): 1468 - 72.
[26] O’HAGAN A. A Markov property for covariance structures [J]. Statistics Research Report, 1998, 98 (13): 510.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Highlights in Science, Engineering and Technology
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
