Article
Engineering, Electrical & Electronic
Ricardo E. Sendrea, Constantinos L. Zekios, Stavros V. Georgakopoulos
Summary: In this work, a new method is proposed for deriving the initial approximate model for a multifidelity surrogate optimization. The proposed method is trained using eigenfunction expansions characterizing the solution domain of the desired geometry and high-fidelity full-wave simulations. The efficiency and effectiveness of the proposed method are demonstrated through studies on various antenna designs, showing significant time savings compared to conventional approaches.
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION
(2023)
Article
Engineering, Mechanical
Kunpeng Li, Yin Liu, Shuo Wang, Xueguan Song
Summary: In this article, a multifidelity surrogate model based on gradient-enhanced radial basis function is proposed to enhance the prediction accuracy of less intensively sampled high-fidelity samples. Through comparisons with other models and investigations on key factors, it is observed that the proposed model demonstrates higher accuracy and lower sensitivity in most cases.
JOURNAL OF MECHANICAL DESIGN
(2021)
Article
Mathematics, Applied
Yiming Xu, Vahid Keshavarzzadeh, Robert M. Kirby, Akil Narayan
Summary: This paper introduces a bandit-learning approach for achieving precise parameter estimation using data of varying fidelities. The proposed algorithm does not require hierarchical model structure or prior knowledge of statistical information, making it efficient and flexible.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2022)
Article
Engineering, Mechanical
Yin Liu, Shuo Wang, Qi Zhou, Liye Lv, Wei Sun, Xueguan Song
Summary: This paper proposes a modified MFS model based on RBF, which can analyze two fidelity information by adaptively obtaining the scale factor. The performance of the model is compared with other MFS models and single-fidelity RBF, and the impact of different high-fidelity sample sizes on prediction accuracy is analyzed. The results show that the proposed model can fully utilize two fidelity sample sets and has good prediction accuracy and robustness.
CHINESE JOURNAL OF MECHANICAL ENGINEERING
(2022)
Article
Computer Science, Interdisciplinary Applications
Vishal Raul, Leifur Leifsson
Summary: This work proposes a multifidelity modeling approach to mitigate adverse characteristics of airfoil dynamic stall through aerodynamic shape optimization (ASO). The approach combines data from high-fidelity and low-fidelity computational fluid dynamics simulations to efficiently determine an optimum airfoil shape. The proposed approach not only delays the dynamic stall angle but also reduces the peak values of the aerodynamic coefficients compared to the baseline airfoil. Moreover, it achieves a better design with less computational time compared to the HF-KR approach.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2023)
Article
Automation & Control Systems
Fangfang Zhang, Yi Mei, Su Nguyen, Mengjie Zhang
Summary: Dynamic flexible job shop scheduling (JSS) has attracted attention for its practical application value, requiring complex routing decisions. Genetic programming (GP) has been successfully applied to evolve scheduling heuristics for JSS. However, simulation-based evaluation is computationally expensive. This article proposes a novel multifidelity-based surrogate-assisted GP to reduce computational cost without sacrificing performance.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Multidisciplinary Sciences
Amelia E. Sancilio, Richard T. D'Aquila, Elizabeth M. McNally, Matt E. Velez, Michael G. Ison, Alexis R. Demonbreun, Thomas W. McDade
Summary: The immunoassay-based method validated in this study can quantify neutralization of the spike-ACE2 interaction in a single drop of capillary whole blood collected as dried blood spot samples. This scalable and minimally invasive alternative to venous blood collection offers a practical way to measure neutralizing antibodies and estimate immunity levels against COVID-19 in the general population.
SCIENTIFIC REPORTS
(2021)
Article
Computer Science, Artificial Intelligence
Shuai Zhang, Kunpeng Li, Shuo Wang, Jianji Li, Yong Pang, Xueguan Song
Summary: In this study, a recursive surrogate model was developed using a generalized regression neural network and variable-fidelity surrogate method. The model continuously improves its prediction accuracy through a novel recursive correction method, resulting in a model with sufficient predictive accuracy. Comparative experiments demonstrated that the proposed model outperforms other benchmark models in terms of predictive accuracy and robustness. This model provides a novel option for engineering design optimization, and the recursive correction method offers a new approach for improving prediction accuracy in other regression models.
APPLIED SOFT COMPUTING
(2023)
Article
Engineering, Chemical
Bianca Williams, Selen Cremaschi
Summary: This study evaluates the performance of eight surrogate modeling techniques in surface approximation and surrogate-based optimization, finding that multivariate adaptive regression spline models and Gaussian process regression provide the most accurate predictions for surface approximation. Additionally, random forests, support vector machine regression, and Gaussian process regression models are the most reliable for identifying optimum locations and values in surrogate-based optimization.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2021)
Article
Engineering, Mechanical
Xiaonan Lai, Xiwang He, Shuo Wang, Xiaobang Wang, Wei Sun, Xueguan Song
Summary: In this study, a lightweight digital twin model was developed to monitor and predict the structural safety of a crane boom in real time, by combining the MFS model and sensor data. The results showed that the lightweight digital twin improved accuracy, reduced computational cost, and analyzed the uncertainty from the physical space to enhance reliability.
JOURNAL OF MECHANICAL DESIGN
(2022)
Article
Mathematics, Interdisciplinary Applications
Baptiste Kerleguer
Summary: This paper presents a surrogate modeling method for a complex numerical code in a multifidelity framework. The method uses a Gaussian process regression approach with experimental design to emulate the high-fidelity code output. The resulting surrogate model shows better performance in predicting the high-fidelity code output compared to standard dimension reduction techniques.
SIAM-ASA JOURNAL ON UNCERTAINTY QUANTIFICATION
(2023)
Article
Engineering, Mechanical
Kunpeng Li, Xiwang He, Liye Lv, Jiaxiang Zhu, Guangbo Hao, Haiyang Li, Xueguan Song
Summary: This paper proposes a single-fidelity surrogate model with a hierarchical structure called Nonlinearity Integrated Correlation Mapping Surrogate (NI-CMS) model, which overcomes the expensive modeling cost by establishing a low-fidelity model and correcting it using the idea of a multi-fidelity surrogate model. The performance of the proposed model is evaluated through numerical test functions and a surrogate-based digital twin experiment, demonstrating its advantages over existing models.
JOURNAL OF MECHANICAL DESIGN
(2023)
Article
Engineering, Multidisciplinary
Daiyu Zhang, Bei Zhang, Zhidong Wang, Xinyao Zhu
Summary: This study proposes an efficient surrogate-based optimization method based on multifidelity model and geometric constraint gradient information, which reduces the calculation times of high-fidelity CFD model and geometric constraints during the shape optimization process of BWBUG, greatly improving the optimization efficiency. The effectiveness and efficiency of the proposed method are verified by performing the shape optimization of a BWBUG and comparing with traditional surrogate-based optimization methods.
MATHEMATICAL PROBLEMS IN ENGINEERING
(2021)
Article
Automation & Control Systems
Jianping Luo, YongFei Dong, Zexuan Zhu, Wenming Cao, Xia Li
Summary: This study proposes a surrogate methodology based on information transfer to improve the estimation effectiveness of surrogate models in multiobjective optimization problems. By mapping related tasks and training a multitask Gaussian process model, the confidence in parameter learning is enhanced, and the predicted values of objective functions can be obtained through reverse mapping. Experimental tests demonstrate that this approach outperforms other surrogate-based optimization algorithms.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Mathematics, Applied
Steffen W. R. Werner, Michael L. Overton, Benjamin Peherstorfer
Summary: In this work, multifidelity gradient sampling methods are introduced to speed up the optimization process by leveraging low-cost, low-fidelity models. Numerical experiments demonstrate significant speedup compared to single-fidelity gradient sampling methods.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2023)