Bayesian Surrogate Learning for Uncertainty Analysis of Coupled Multidisciplinary Systems

Engineering systems are often composed of many subsystems that interact with each other. These subsystems, referred to as disciplines, contain many types of uncertainty and in many cases are feedback-coupled with each other. In designing these complex systems, one needs to assess the stationary behavior of these systems for the sake of stability and reliability. This requires the system level uncertainty analysis of the multidisciplinary systems, which is often computationally intractable. To overcome this issue, techniques have been developed for capturing the stationary behavior of the coupled multidisciplinary systems through available data of individual disciplines. The accuracy and convergence of the existing techniques depend on a large amount of data from all disciplines, which are not available in many practical problems. Toward this, we have developed an adaptive methodology that adds the minimum possible number of samples from individual disciplines to achieve an accurate and reliable uncertainty propagation in coupled multidisciplinary systems. The proposed method models each discipline function via Gaussian process (GP) regression to derive a closed-form policy. This policy sequentially selects a new sample point that results in the highest uncertainty reduction over the distribution of the coupling design variables. The effectiveness of the proposed method is demonstrated in the uncertainty analysis of an aerostructural system and a coupled numerical example.

[1]  A. O'Hagan,et al.  Bayesian calibration of computer models , 2001 .

[2]  Roger Ghanem,et al.  Dimension reduction in stochastic modeling of coupled problems , 2011, 1112.4761.

[3]  Sankaran Mahadevan,et al.  Decoupled approach to multidisciplinary design optimization under uncertainty , 2007 .

[4]  Sankaran Mahadevan,et al.  Likelihood-Based Approach to Multidisciplinary Analysis Under Uncertainty , 2012 .

[5]  Sankaran Mahadevan,et al.  Robustness-Based Design Optimization of Multidisciplinary System Under Epistemic Uncertainty , 2013 .

[6]  Seyede Fatemeh Ghoreishi,et al.  Bayesian Optimization Objective-Based Experimental Design , 2020, 2020 American Control Conference (ACC).

[7]  Douglas Allaire,et al.  Gaussian Process Regression for Bayesian Fusion of Multi-Fidelity Information Sources , 2018 .

[8]  Wei Chen,et al.  Collaborative Reliability Analysis under the Framework of Multidisciplinary Systems Design , 2005 .

[9]  Donald Geman,et al.  Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[10]  Glen Mullineux,et al.  Introducing Loading Uncertainty in Topology Optimization , 2009 .

[11]  Anirban Chaudhuri,et al.  Multifidelity Uncertainty Propagation in Coupled Multidisciplinary Systems , 2016 .

[12]  Douglas Allaire,et al.  Adaptive Uncertainty Propagation for Coupled Multidisciplinary Systems , 2017 .

[13]  Seyede Fatemeh Ghoreishi,et al.  Bayesian Optimization for Efficient Design of Uncertain Coupled Multidisciplinary Systems , 2020, 2020 American Control Conference (ACC).

[14]  B. Carlin,et al.  On the Convergence of Successive Substitution Sampling , 1992 .

[15]  Sankaran Mahadevan,et al.  Stochastic Multidisciplinary Analysis with High-Dimensional Coupling , 2016 .

[16]  John E. Renaud,et al.  Implicit Uncertainty Propagation for Robust Collaborative Optimization , 2006 .

[17]  Karen Willcox,et al.  A decomposition‐based approach to uncertainty analysis of feed‐forward multicomponent systems , 2014 .

[18]  Jaroslaw Sobieszczanskisobieski,et al.  On the sensitivity of complex, internally coupled systems , 1988 .

[19]  Douglas Allaire,et al.  Compositional Uncertainty Analysis via Importance Weighted Gibbs Sampling for Coupled Multidisciplinary Systems , 2016 .

[20]  Natasha Smith,et al.  Efficient first-order reliability analysis of multidisciplinary systems , 2006 .

[21]  Seyede Fatemeh Ghoreishi,et al.  Bayesian Control of Large MDPs with Unknown Dynamics in Data-Poor Environments , 2018, NeurIPS.

[22]  Douglas L. Allaire,et al.  MFBO-SSM: Multi-Fidelity Bayesian Optimization for Fast Inference in State-Space Models , 2019, AAAI.

[23]  Yoshua Bengio,et al.  Entropy Regularization , 2006, Semi-Supervised Learning.

[24]  Roger Ghanem,et al.  Reduced chaos expansions with random coefficientsin reduced‐dimensional stochastic modeling of coupled problems , 2012, 1207.0910.

[25]  Kenneth T. Moore,et al.  Optimization of Turbine Engine Cycle Analysis with Analytic Derivatives , 2016 .

[26]  Jack J. McNamara,et al.  Studies on Fluid-Thermal-Structural Coupling for Aerothermoelasticity in Hypersonic Flow , 2010 .

[27]  Seyede Fatemeh Ghoreishi,et al.  Optimal Finite-Horizon Perturbation Policy for Inference of Gene Regulatory Networks , 2021, IEEE Intelligent Systems.

[28]  Paul G. Constantine,et al.  Efficient uncertainty propagation for network multiphysics systems , 2013, 1308.6520.

[29]  Yoshua Bengio,et al.  Semi-supervised Learning by Entropy Minimization , 2004, CAP.

[30]  R. A. Leibler,et al.  On Information and Sufficiency , 1951 .

[31]  Xiaoping Du,et al.  Efficient Uncertainty Analysis Methods for Multidisciplinary Robust Design , 2002 .

[32]  Michel van Tooren,et al.  Review of uncertainty-based multidisciplinary design optimization methods for aerospace vehicles , 2011 .

[33]  Karen Willcox,et al.  Multifidelity Uncertainty Propagation via Adaptive Surrogates in Coupled Multidisciplinary Systems , 2018 .

[34]  Panos Y. Papalambros,et al.  Design Optimization of Hierarchically Decomposed Multilevel Systems Under Uncertainty , 2006 .

[35]  Karen Willcox,et al.  A Decomposition Approach to Uncertainty Analysis of Multidisciplinary Systems , 2012 .