Scholarlink

<p>The Kriging model often suffers from the curse of dimensionality. To address this issue, the authors propose a method to enhance the accuracy of high-dimensional models and accelerate computational efficiency. The core innovation of the method lies in analyzing the correlations between inputs and outputs using the Hilbert–Schmidt Independence Criterion, through which correlation coefficients are obtained. These coefficients are then stratified via the K-means clustering method, with the Bayesian Information Criterion employed to determine the optimal number of layers. Based on this stratification, a layered low-dimensional kernel function is constructed to establish the Kriging model. To evaluate the accuracy and robustness of our method, this paper tests it using multiple mathematical functions with inputs ranging from 40-D to 80-D and a complex engineering case with 49-D inputs. The results show that, compared with existing methods, the proposed method achieves superior performance in both accuracy and stability. Furthermore, for high-dimensional, strongly nonlinear, and small-samples engineering problems, this method exhibits a higher accuracy ceiling and stronger competitiveness.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A new Kriging model for high-dimensional problems based on correlation analysis and adaptive stratification

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Pengzhi Chen,
  • Linyong Chen,
  • Fan Yang,
  • Xufeng Yang

摘要

The Kriging model often suffers from the curse of dimensionality. To address this issue, the authors propose a method to enhance the accuracy of high-dimensional models and accelerate computational efficiency. The core innovation of the method lies in analyzing the correlations between inputs and outputs using the Hilbert–Schmidt Independence Criterion, through which correlation coefficients are obtained. These coefficients are then stratified via the K-means clustering method, with the Bayesian Information Criterion employed to determine the optimal number of layers. Based on this stratification, a layered low-dimensional kernel function is constructed to establish the Kriging model. To evaluate the accuracy and robustness of our method, this paper tests it using multiple mathematical functions with inputs ranging from 40-D to 80-D and a complex engineering case with 49-D inputs. The results show that, compared with existing methods, the proposed method achieves superior performance in both accuracy and stability. Furthermore, for high-dimensional, strongly nonlinear, and small-samples engineering problems, this method exhibits a higher accuracy ceiling and stronger competitiveness.