Multi-Model Bayesian Optimization for Simulation-Based Design (DETC2020-22651)

2021 ◽  
pp. 1-39
Author(s):  
Siyu Tao ◽  
Anton van Beek ◽  
Daniel Apley ◽  
Wei Chen
Keyword(s):  
Simulation Models ◽  
Global Optimum ◽  
Bayesian Optimization ◽  
System Response ◽  
Engineering Systems ◽  
Simulation Based Design ◽  
Simulation Based ◽  
Need To Evaluate ◽  
Complex Engineering ◽  
Component Models

Abstract We enhance the Bayesian optimization (BO) approach for simulation-based design of engineering systems consisting of multiple interconnected expensive simulation models. The goal is to find the global optimum design with minimal model evaluation costs. A commonly used approach is to treat the whole system as a single expensive model and apply an existing BO algorithm. This approach is inefficient due to the need to evaluate all the component models in each iteration. We propose a multi-model BO approach that dynamically and selectively evaluates one component model per iteration based on the uncertainty quantification of linked emulators (metamodels) and the knowledge gradient of system response as the acquisition function. Building on our basic formulation, we further solve problems with constraints and feedback couplings that often occur in real complex engineering design by penalizing the objective emulator and reformulating the original problem into a decoupled one. The superior efficiency of our approach is demonstrated through solving two analytical problems and the design optimization of a multidisciplinary electronic packaging system.

Bayesian Optimization for Simulation-Based Design of Multi-Model Systems

2020 ◽  
Author(s):  
Siyu Tao ◽  
Anton van Beek ◽  
Daniel W. Apley ◽  
Wei Chen
Keyword(s):  
Simulation Models ◽  
Model Systems ◽  
Bayesian Optimization ◽  
Simulation Based Design ◽  
Simulation Based ◽  
Need To Evaluate ◽  
Complex Engineering ◽  
Component Models ◽  
Knowledge Gradient ◽  
Expensive Model

Abstract We address the problem of simulation-based design using multiple interconnected expensive simulation models, each modeling a different subsystem. Our goal is to find the globally optimal design with minimal model evaluation costs. To our knowledge, the best existing approach is to treat the whole system as a single expensive model and apply an existing Bayesian optimization (BO) algorithm. This approach is likely inefficient due to the need to evaluate all the component models in each iteration. We propose a multi-model BO approach that dynamically and selectively evaluates one component model per iteration based on linked emulators for uncertainty quantification and the system knowledge gradient (KG) as acquisition function. Building on this, we resolve problems with constraints and feedback couplings that often occur in real complex engineering design by penalizing the objective emulator and reformulating the original problem into a decoupled one. The superior efficiency of our approach is demonstrated through solving an analytical problem and a multidisciplinary design problem of electronic packaging optimization.

Analytical Variance-Based Global Sensitivity Analysis in Simulation-Based Design Under Uncertainty

2004 ◽  
Author(s):  
Wei Chen ◽  
Ruichen Jin ◽  
Agus Sudjianto
Keyword(s):  
Sensitivity Analysis ◽  
Global Sensitivity Analysis ◽  
Simulation Models ◽  
Design Under Uncertainty ◽  
Global Sensitivity ◽  
Functional Relationships ◽  
Simulation Based Design ◽  
Simulation Based ◽  
Input Variables ◽  
The Impact

The importance of sensitivity analysis in engineering design cannot be over-emphasized. In design under uncertainty, sensitivity analysis is performed with respect to the probabilistic characteristics. Global sensitivity analysis (GSA), in particular, is used to study the impact of variations in input variables on the variation of a model output. One of the most challenging issues for GSA is the intensive computational demand for assessing the impact of probabilistic variations. Existing variance-based GSA methods are developed for general functional relationships but require a large number of samples. In this work, we develop an efficient and accurate approach to GSA that employs analytic formulations derived from metamodels of engineering simulation models. We examine the types of GSA needed for design under uncertainty and derive generalized analytical formulations of GSA based on a variety of metamodels commonly used in engineering applications. The benefits of our proposed techniques are demonstrated and verified through both illustrative mathematical examples and the robust design for improving vehicle handling performance.

Confidence Estimates due to Uncertainty in Multi-Disciplinary Computational Analysis

2013 ◽  
Author(s):  
Animesh Dey ◽  
Robert Tryon
Keyword(s):  
Computational Analysis ◽  
Input Parameter ◽  
Virtual Simulation ◽  
System Response ◽  
Model Refinement ◽  
Simulation Based Design ◽  
Simulation Based ◽  
Final Output ◽  
Input Variables ◽  
Confidence Estimates

Simulation-based design and certification is fundamentally about making decisions with uncertainty. However, minimizing uncertainty comes at a price — more testing to better define the variability in input parameters, higher fidelity analyses at a finer scale to limit the uncertainty in the physics, etc. Variability in each input parameter does not affect the uncertainty in the system response equally. Nor does every model refinement reduce the uncertainty in the system response. This paper presents a computational methodology that estimates the sensitivity of uncertainty in input variables and the sensitivity of modeling approximations to the final output. In the current age of large multi-disciplinary virtual simulation, this is useful in determining how to minimize overall uncertainty in analytical predictions. In addition, the methodology can be used to optimize for the best use of computational and testing resources to arrive at most robust predictions.

Modeling and Simulation Methods for Design of Engineering Systems

2000 ◽  
Vol 1 (1) ◽  
pp. 84-91 ◽  
Author(s):  
Rajarishi Sinha ◽  
Christiaan J. J. Paredis ◽  
Vei-Chung Liang ◽  
Pradeep K. Khosla
Keyword(s):  
Modeling And Simulation ◽  
Design Process ◽  
State Of The Art ◽  
Simulation Software ◽  
Modeling Languages ◽  
Simulation Methods ◽  
Design Tools ◽  
Engineering Systems ◽  
Simulation Based Design ◽  
Simulation Based

This article presents an overview of the state-of-the art in modeling and simulation, and studies to which extent current simulation technologies can effectively support the design process. For simulation-based design, modeling languages and simulation environments must take into account the special characteristics of the design process. For instance, languages should allow models to be easily updated and extended to accommodate the various analyses performed throughout the design process. Furthermore, the simulation software should be well integrated with the design tools so that designers and analysts with expertise in different domains can effectively collaborate on the design of complex artifacts. This review focuses in particular on modeling for design of multi-disciplinary engineering systems that combine continuous time and discrete time phenomena.

Applying Bayesian Optimization for Calibration of Transportation Simulation Models

2020 ◽  
Vol 2674 (10) ◽  
pp. 215-228
Author(s):  
Di Sha ◽  
Kaan Ozbay ◽  
Yue Ding
Keyword(s):  
Simulation Model ◽  
Simulation Models ◽  
Global Optimum ◽  
Bayesian Optimization ◽  
Local Optimum ◽  
Computationally Efficient ◽  
Actual System ◽  
Transportation Simulation ◽  
Calibration Accuracy ◽  
Comparative Results

The parameters of a transportation simulation model need to pass through a careful calibration process to ensure that the model’s output is as close as possible to the actual system. Owing to the computationally expensive and black-box nature of a simulation model, there is a need for robust and efficient calibration algorithms. This paper proposes a Bayesian optimization framework for the high-dimensional calibration problem of transportation simulation models. Bayesian optimization uses acquisition functions to determine more promising values for future evaluation, instead of relying on local gradient approximations. It guarantees convergence to the global optimum with a reduced number of evaluations, therefore is very computationally efficient. The proposed algorithm is applied to the calibration of a simulation network coded in simulation of urban mobility (SUMO), an open-source microscopic transportation simulation platform, and compared with a well-known method named simultaneous perturbation stochastic approximation (SPSA). To assess the calibration accuracy, speed distributions obtained from the two models calibrated using these two different methods are compared with the observation. For both the Bayesian optimization and SPSA results, the simulated and observed distributions are validated to be from the same distribution at a 95% confidence level for multiple sensor locations. Thus, the calibration accuracy of the two approaches are both acceptable for a stochastic transportation simulation model. However, Bayesian optimization shows a better convergence and a higher computational efficiency than SPSA. In addition, the comparative results of multiple implementations validate its robustness for a noisy objective function, unlike SPSA which may sometimes get stuck in a local optimum and fail to converge in a global solution.

Powertrain Modeling With Modelica

2000 ◽  
Author(s):  
Michael Tiller ◽  
Cleon Davis ◽  
Hubertus Tummescheit ◽  
Nizar Trigui
Keyword(s):  
Mechanical Response ◽  
Analysis Tool ◽  
Engineering Systems ◽  
Behavioral Models ◽  
Combustion Gas ◽  
Computational Performance ◽  
Complex Engineering ◽  
Component Models ◽  
Complex Engineering Systems ◽  
Traditional Component

Abstract In this paper we will describe the development of models for prediction of powertrain performance. Our goal is to develop a library of components to model combustion, gas dynamics and mechanical response. We will also demonstrate the ease with which we can replace traditional component models (e.g., mechanically actuated valves) with non-traditional component models (e.g., electro-mechanically actuated valves) without having to change or reformulate any of the other components in our system. The models were developed using the Modelica modeling language (Modelica Design Group, 1999) which allows component-based descriptions of behavior for complex engineering systems. Modelica is particularly well suited for creating behavioral models that are typical for powertrain plant models In addition to writing component models, the freely available Modelica Standard Library contains basic models from various engineering disciplines (e.g., resistors, shafts, springs). With this approach, models can be symbolically preprocessed to improve computational performance. In addition, code can be generated from the Modelica model which can be used as either a stand alone analysis tool, imported into Simulink as an S-function or downloaded for use in real-time hardware in the loop experiments.

Scalable Objective-Driven Batch Sampling in Simulation-Based Design for Models With Heteroscedastic Noise

2020 ◽  
Author(s):  
Anton van Beek ◽  
Umar Farooq Ghumman ◽  
Joydeep Munshi ◽  
Siyu Tao ◽  
TeYu Chien ◽  
...  
Keyword(s):  
Stochastic Simulation ◽  
Adaptive Sampling ◽  
Numerical Test ◽  
Computational Cost ◽  
Simulation Models ◽  
Global Optimum ◽  
Bayesian Optimization ◽  
Sampling Scheme ◽  
Sampling Location ◽  
Sampling Locations

Abstract Objective-driven adaptive sampling is a widely used tool for the optimization of deterministic black-box functions. However, the optimization of stochastic simulation models as found in the engineering, biological, and social sciences is still an elusive task. In this work, we propose a scalable adaptive batch sampling scheme for the optimization of stochastic simulation models with input-dependent noise. The developed algorithm has two primary advantages: (i) by recommending sampling batches, the designer can benefit from parallel computing capabilities, and (ii) by replicating of previously observed sampling locations the method can be scaled to higher-dimensional and more noisy functions. Replication improves numerical tractability as the computational cost of Bayesian optimization methods is known to grow cubicly with the number of unique sampling locations. Deciding when to replicate and when to explore depends on what alternative minimizes the posterior prediction accuracy at and around the spatial locations expected to contain the global optimum. The algorithm explores a new sampling location to reduce the interpolation uncertainty and replicates to improve the accuracy of the mean prediction at a single sampling location. Through the application of the proposed sampling scheme to two numerical test functions and one real engineering problem, we show that we can reliably and efficiently find the global optimum of stochastic simulation models with input-dependent noise.

Bayesian Optimization for Simulation-Based Design of Multi-Model Systems

2021 ◽  
Author(s):  
Siyu Tao ◽  
Anton Van Beek ◽  
Daniel W. Apley ◽  
Wei Chen
Keyword(s):  
Model Systems ◽  
Bayesian Optimization ◽  
Simulation Based Design ◽  
Simulation Based

A Two-Level Product Model for Simulation-Based Design of Mechanical Systems

1997 ◽  
Author(s):  
Kuang-Hua Chang ◽  
Kyung K. Choi ◽  
Jeff J. Y. Wang ◽  
Chung-Shin Tsai ◽  
Edwin Hardee
Keyword(s):  
Simulation Models ◽  
Design Stage ◽  
Physical Parameters ◽  
Cad Model ◽  
Preliminary Design ◽  
Product Model ◽  
Detailed Design ◽  
Simulation Based Design ◽  
Simulation Based ◽  
Primary Focus

Abstract This paper presents a two-level product modeling method that supports Simulation-Based Design (SBD) of mechanical systems, primarily ground vehicles and heavy equipment, for preliminary and detailed design. A Computer-Aided Design (CAD) model combined with engineering parameters and mathematical equations that describe physical behavior of the mechanical system constitute its product model for SBD. For preliminary design, improvement of system performance, including dynamics and human factors, is the primary focus. A CAD model with reasonably accurate physical parameters, such as mass properties, is defined as the base definition of the product model. A parametric study can be conducted to search for design alternatives using dimension parameters created in the parameterized CAD model. Component designs are the primary focus in the detailed design stage. A detailed product model is evolved from that of the preliminary design, by refining geometric representation of mechanical components in CAD and expanding product assembly into parts and sub-assemblies for further engineering analysis. In the detailed design stage, a systematic design trade-off method is usually needed for design improvement. In both design stages, CAD and Computer-Aided Engineering (CAE) mappings that tie dimension parameters in the CAD model and physical parameters of simulation models facilitate the parametric study and design trade-off by quickly generating simulation models to simulate performance of the modified design. A High Mobility Multi-Purpose Wheeled Vehicle (HMMWV) is employed to illustrate and demonstrate the modeling method.

The research of the decomposition‐coordination method of multidisciplinary collaboration design optimization

2005 ◽  
Vol 22 (3) ◽  
pp. 274-285 ◽  
Author(s):  
Jianjiang Chen ◽  
Yifang Zhong ◽  
Renbin Xiao ◽  
Jianxun Sun
Keyword(s):  
Design Optimization ◽  
Large Scale ◽  
Global Optimum ◽  
Engineering Problem ◽  
Engineering Systems ◽  
Decomposition Approach ◽  
Content Type ◽  
Complex Design ◽  
Complex Engineering ◽  
Artificial Neural Network Ann

PurposeTo obtain the global optimum of large‐scale complex engineering systems, the paper proposes a decomposition‐coordination method of multidisciplinary design optimization (MDO).Design/methodology/approachA rational decomposition approach based on artificial neural network (ANN) and genetic algorithms is proposed for partitioning the complex design problem into smaller, more tractable subsystems. Once the problem is decomposed into subsystems, each subsystem may be solved in parallel provided that there is some mechanism to coordinate the solutions in the different subsystems. So the response surface approximation model based on the ANN as a coordination method is described and a MDO framework is presented.FindingsThe proposed method was implemented in the design of a tactical missile. Numerical results show the effectiveness of the decomposition‐coordination method, as indicated by both better performance and lower computational requirements.Originality/valueThis paper adopts a novel MDO method to solve complex engineering problem and offers a potential and efficient MDO framework to researchers.

Sign in / Sign up

Export Citation Format

Share Document