Abstract
In product development, numerous design decisions have to be made. Multi-domain virtual prototyping provides a variety of tools to assess technical feasibility of design options, however often requires substantial computational effort for just a single evaluation. A special challenge is therefore the optimal design of product families, which consist of a group of products derived from a common platform. Finding an optimal platform configuration (stating what is shared and what is individually designed for each product) and an optimal design of all products simultaneously leads to a mixed-integer nonlinear black-box optimization model. We present an optimization approach based on metamodels and a metaheuristic. To increase computational efficiency and solution quality, we compare different types of Gaussian process regression metamodels adapted from the domain of machine learning, and combine them with a genetic algorithm. We illustrate our approach on the example of a product family of electrical drives, and investigate the trade-off between solution quality and computational overhead.
Keywords
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
In all cases, an anisotropic squared exponential kernel and a Gaussian likelihood function with a noise level fixed to 10−10 are used. Hyper parameters θ of the model are chosen by maximization of the log likelihood.
References
Fujita, K., Yoshida, H.: Product variety optimization: simultaneous optimization of module combination and module attributes. In: Proceedings of DETC’01, 2001 ASME Design Engineering Technical Conferences and Computers and Information in Engineering Conference (2001)
Simpson, T.W., et al. (eds.).: Advances in Product Family and Product Platform Design. Springer, New York (2014)
Khajavirad, A., et al.: An efficient decomposed multiobjective genetic algorithm for solving the joint product platform selection and product family design problem with generalized commonality. Struct. Multidiscip. Optim. 39(2), 187–201 (2009)
Müller, J.: MISO: mixed-integer surrogate optimization framework. Optim. Eng. 17(1), 177–203 (2016)
Müller, J.: SOCEMO: surrogate optimization of computationally expensive multiobjective problems. INFORMS J. Comput. 29(4), 581–596 (2017)
Stenger, D., et al.: Product Family Design Optimization Using Model-Based Engineering Techniques. Proceedings of the Conference on Operations Research 2017, pp. 495–502. Springer, Cham (2018)
Krige, D.G.: A statistical approach to some basic mine valuation problems on the Witwatersrand. J. South. Afr. Inst. Min. Metall. 52(6), 119–139 (1951)
Rasmussen, C.E., Williams, C.K.I.: Gaussian Processes for Machine Learning. MIT Press, Cambridge, MA (2006). ISBN 026218253X
Forrester, A.I., Keane, A.J.: Recent advances in surrogate-based optimization. Prog. Aerosp. Sci. 45(1–3), 50–79 (2009)
Snelson, E., Ghahramani, Z.: Sparse Gaussian processes using pseudo-inputs. In: Advances in Neural Information Processing Systems, pp. 1257–1264 (2006)
Couckuyt, I., et al.: Blind Kriging: implementation and performance analysis. Adv. Eng. Softw. 49, 1–13 (2012)
Rasmussen, C.E., Nickisch, H.: Gaussian processes for machine learning (GPML) toolbox. J. Mach. Learn. Res. 11, 3011–3015 (2010)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Stenger, D., Altherr, L.C., Abel, D. (2019). Machine Learning and Metaheuristics for Black-Box Optimization of Product Families: A Case-Study Investigating Solution Quality vs. Computational Overhead. In: Fortz, B., Labbé, M. (eds) Operations Research Proceedings 2018. Operations Research Proceedings. Springer, Cham. https://doi.org/10.1007/978-3-030-18500-8_47
Download citation
DOI: https://doi.org/10.1007/978-3-030-18500-8_47
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-18499-5
Online ISBN: 978-3-030-18500-8
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)