Zobraziť v EDS

Research on the Improvement Strategy of Initial Sampling Point Selection in Bayesian Optimization-Based Uncertainty Analysis Method.

Uložené v:
Podrobná bibliografia
Názov: Research on the Improvement Strategy of Initial Sampling Point Selection in Bayesian Optimization-Based Uncertainty Analysis Method.
Autori: Bai, Jinjun, Ji, Xiangrui, Liu, Qing, Song, Yujia, Zheng, Zhongjiu
Zdroj: Progress in Electromagnetics Research C; 2025, Vol. 158, p179-186, 8p
Predmety: ELECTROMAGNETIC compatibility, SAMPLING (Process), ADAPTIVE sampling (Statistics), GAUSSIAN processes, UNCERTAINTY (Information theory), STOCHASTIC programming, REDUCED-order models
Abstrakt: In recent years, uncertainty analysis methods have become a research hotspot in the field of Electromagnetic Compatibility (EMC), and non-intrusive uncertainty analysis methods are widely used in the field of EMC due to their advantages such as easy solver generalization and easy programming. The proposal of Bayesian optimization-based uncertainty analysis method further enhances the competitiveness of non-intrusive uncertainty analysis methods in solving complex EMC simulation problems. However, in traditional Bayesian optimization-based uncertainty analysis methods, Latin hypercube sampling strategy is used to construct the initial Gaussian process model, which lacks adaptive adjustment capability, and the quality of the initial Gaussian process model has a significant impact on the efficiency of subsequent calculations and the accuracy of the final results. This defect limits the computational efficiency and accuracy of Bayesian optimization methods in uncertainty analysis applications. In response to this issue, this paper proposes an active sampling strategy based on the Stochastic Reduced Order Model (SROM) method. This strategy improves the fitness function used by the SROM method in clustering to enhance the representativeness of the training set to the sampling space. By using this active sampling strategy instead of Latin hypercube sampling strategy, a higher quality initial Gaussian process model can be constructed, and the accuracy of Bayesian optimization method uncertainty analysis calculation is improved in the example, verifying the effectiveness of the proposed initial sampling point selection improvement strategy. [ABSTRACT FROM AUTHOR]
Copyright of Progress in Electromagnetics Research C is the property of Electromagnetics Academy and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Databáza: Complementary Index
Popis
Abstrakt:In recent years, uncertainty analysis methods have become a research hotspot in the field of Electromagnetic Compatibility (EMC), and non-intrusive uncertainty analysis methods are widely used in the field of EMC due to their advantages such as easy solver generalization and easy programming. The proposal of Bayesian optimization-based uncertainty analysis method further enhances the competitiveness of non-intrusive uncertainty analysis methods in solving complex EMC simulation problems. However, in traditional Bayesian optimization-based uncertainty analysis methods, Latin hypercube sampling strategy is used to construct the initial Gaussian process model, which lacks adaptive adjustment capability, and the quality of the initial Gaussian process model has a significant impact on the efficiency of subsequent calculations and the accuracy of the final results. This defect limits the computational efficiency and accuracy of Bayesian optimization methods in uncertainty analysis applications. In response to this issue, this paper proposes an active sampling strategy based on the Stochastic Reduced Order Model (SROM) method. This strategy improves the fitness function used by the SROM method in clustering to enhance the representativeness of the training set to the sampling space. By using this active sampling strategy instead of Latin hypercube sampling strategy, a higher quality initial Gaussian process model can be constructed, and the accuracy of Bayesian optimization method uncertainty analysis calculation is improved in the example, verifying the effectiveness of the proposed initial sampling point selection improvement strategy. [ABSTRACT FROM AUTHOR]
ISSN:19378718
DOI:10.2528/PIERC25051404