Efficient Use of Partially Converged Simulations in Evolutionary Optimization

Branke, Juergen; Asafuddoula, Md.; Bhattacharjee, Kalyan Shankar; Ray, Tapabrata

doi:10.1109/tevc.2016.2569018

Cited by 28 publications

(12 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in aerodynamic optimization, Navier-Stokes equations or coupled boundary layer methods can be adopted, and for both methods, either 2-D or 3-D simulations can be carried out [13]. In addition, the resolution of the simulations can be adjusted by changing the resolution of the mesh size [32], or by controlling the tolerance error [4] or the maximum number of iterations [2] in CFD simulations. Therefore, the number of fidelity levels in multifidelity simulations could range from two to infinity, depending on the specific application.…”

Section: Multi-fidelity Simulationsmentioning

confidence: 99%

“…Resolution errors can also result from partial or incomplete simulations. For example, in CFD simulations, usually a large number of iterations are needed to accurately solve differential equations [2]. In Fig.…”

Section: A Approximation Errors In Multi-fidelity Optimizationmentioning

confidence: 99%

“…Such numerical simulations are usually computationally expensive, yet the computational complexity of the simulations can be tuned. For example, the complexity of computational fluid dynamics (CFD) simulations can be tuned by changing the number of iterations [2], [3], [4], [5]. Simulationdriven optimization where the fidelity of the simulations can be controlled are known as multi-fidelity optimization (also variable fidelity optimization) [6], which has been commonly used in design of microwave structures [7], aerodynamic optimization [8], [9], ship design [10], engine design [11], and rocket design [12].…”

Section: Introductionmentioning

confidence: 99%

“…Unfortunately, the two functions used for representing two fidelity levels do not consider the correlation between fitness evaluations of different fidelity levels. The test problem recently reported in [2] considers partially converged simulations and has six fidelity levels. In that work, the correlation between different fidelity levels is taken into account, although these test problems are not scalable, nor easily extendible to other situations where various levels of fidelity exist.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Generic Test Suite for Evolutionary Multifidelity Optimization

Wang

Jin

Doherty

2018

IEEE Trans. Evol. Computat.

View full text Add to dashboard Cite

Many real-world optimization problems involve computationally intensive numerical simulations to accurately evaluate the quality of solutions. Usually the fidelity of the simulations can be controlled using certain parameters and there is a trade-off between simulation fidelity and computational cost, i.e., the higher the fidelity, the more complex the simulation will be. To reduce the computational time in simulation-driven optimization, it is a common practice to use multiple fidelity levels in search for the optimal solution. So far, not much work has been done in evolutionary optimization that considers multiple fidelity levels in fitness evaluations. In this work, we aim to develop test suites that are able to capture some important characteristics in real-world multi-fidelity optimization, thereby offering a useful benchmark for developing evolutionary algorithms for multi-fidelity optimization. To demonstrate the usefulness of the proposed test suite, three strategies for adapting the fidelity level of the test problems during optimization are suggested and embedded in a particle swarm optimization algorithm. Our simulation results indicate that the use of changing fidelity is able to enhance the performance and reduce the computational cost of the particle swarm optimization, which is desired in solving expensive optimization problems.

show abstract

Section: Multi-fidelity Simulationsmentioning

confidence: 99%

Section: A Approximation Errors In Multi-fidelity Optimizationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Generic Test Suite for Evolutionary Multifidelity Optimization

Wang

Jin

Doherty

2018

IEEE Trans. Evol. Computat.

View full text Add to dashboard Cite

show abstract

“…On the other hand, the iteration for the partially converged simulation to be used as the LF simulations is set to 100, which means that the computational cost ratio between the LF and HF simulation is 0.2 for this case. This fully and partially converged simulation strategy has been employed for optimization [83,84], surrogate model construction [85], and UQ [71]. One advantage of this strategy is that some LF samples are evaluated together with the HF samples, which leads to further reduction in the computational cost.…”

Section: Transonic Airfoil In Inviscid Flowmentioning

confidence: 99%

Global sensitivity analysis via multi-fidelity polynomial chaos expansion

Palar

Zuhal

Shimoyama

et al. 2018

Reliability Engineering & System Safety

View full text Add to dashboard Cite

The presence of uncertainties are inevitable in engineering design and analysis, where failure in understanding their effects might lead to the structural or functional failure of the systems. The role of global sensitivity analysis in this aspect is to quantify and rank the effects of input random variables and their combinations to the variance of the random output. In problems where the use of expensive computer simulations are required, metamodels are widely used to speed up the process of global sensitivity analysis. In this paper, a multi-fidelity framework for global sensitivity analysis using polynomial chaos expansion (PCE) is presented. The goal is to accelerate the computation of Sobol sensitivity indices when the deterministic simulation is expensive and simulations with multiple levels of fidelity are available. This is especially useful in cases where a partial differential equation solver computer code is utilized to solve engineering problems. The multi-fidelity PCE is constructed by combining the lowfidelity and correction PCE. Following this step, the Sobol indices are computed using this combined PCE. The PCE coefficients for both low-fidelity and correction PCE are computed with spectral projection technique and sparse grid integration. In order to demonstrate the capability of the proposed method for sensitivity analysis, several simulations are conducted. On the aerodynamic example, the multi-fidelity approach is able to obtain an accurate value of Sobol indices with 36.66% computational cost compared to the standard single-fidelity PCE for a nearly similar accuracy.

show abstract

Active Robust Optimization: Optimizing for Robustness of Changeable Products

Salomon

2019

Springer Theses

View full text Add to dashboard Cite

To succeed in a demanding and competitive market, great attention needs to be given to the process of product design. Incorporating optimization into the process enables the designer to find high-quality products according to their simulated performance. However, the actual performance may differ from the simulation results due to a variety of uncertainty factors. Robust optimization is commonly used to search for products that are less affected by the anticipated uncertainties. Changeability can improve the robustness of a product, as it allows the product to be adapted to a new configuration whenever the uncertain conditions change. This ability provides the changeable product with an active form of robustness.Several methodologies exist for engineering design of changeable products, none of which includes optimization. This study presents the Active Robust Optimization (ARO) framework that offers the missing tools for optimizing changeable products. A new optimization problem is formulated, named Active Robust Optimization Problem (AROP). The benefit in designing solutions by solving an AROP lies in the realistic manner adaptation is considered when assessing the solutions' performance.The novel methodology can be applied to optimize any product that can be classified as a changeable product, i.e., it can be adjusted by its user during normal operation. This definition applies to a huge variety of applications, ranging from simple products such as fans and heaters, to complex systems such as production halls and transportation systems. The ARO framework is described in this dissertation and its unique features are studied. Its ability to find robust changeable solutions is examined for different sources of uncertainty, robustness criteria and sampling conditions. Additionally, a framework for Active Robust Multi-objective Optimization is developed. This generalisation of ARO itself presents many challenges, not encountered in previous studies. Novel approaches for evaluating and comparing changeable designs comprising multiple objectives are proposed along with algorithms for solving multi-objective AROPs.The framework and associated methodologies are demonstrated on two applications from different fields in engineering design. The first is an adjustable optical table, and the second is the selection of gears in a gearbox.iii Statement of OriginalityUnless otherwise stated in the text, the work described in this thesis was carried out solely by the candidate. None of this work has already been

show abstract

Efficient Use of Partially Converged Simulations in Evolutionary Optimization

Cited by 28 publications

References 36 publications

A Generic Test Suite for Evolutionary Multifidelity Optimization

A Generic Test Suite for Evolutionary Multifidelity Optimization

Global sensitivity analysis via multi-fidelity polynomial chaos expansion

Active Robust Optimization: Optimizing for Robustness of Changeable Products

Contact Info

Product

Resources

About