Modified Matrix Decomposition Method for Generating Multivariate Three-Dimensional Cross-Correlated Random Fields
文献类型：期刊论文
| 作者 | Tian, Ning; Chen, Jian; Wang, Zhansheng; Tian, Kaiwei; Huang, Juehao |
| 刊名 | ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
 |
| 出版日期 | 2023-05-21 |
| 关键词 | Soil heterogeneity Spatial variability Modified matrix decomposition Multivariate random field Cross-correlation structure |
| ISSN号 | 2193-567X |
| 英文摘要 | It is essential to generate multivariate three-dimensional cross-correlated random fields since the spatial variability of geomaterials has a great influence on the failure mechanism and reliability of geotechnical structures. Standard matrix decomposition (MD) method has now been commonly employed to generate random field due to its simplicity. However, the traditional MD method has a dilemma that its cross-correlated matrix is tending to ignore the influence of spatial location between different spatial points when applied to generate multivariate cross-correlated random field. Moreover, the generation efficiency will be low due to computational issues associated with large-dimensional correlation matrices operations when generating a three-dimensional random field. Starting from the correlation of parameters, this paper develops a new mathematical inversion model to generate multivariate three-dimensional cross-correlated random fields with extended cross-correlation matrix, namely modified matrix decomposition (M-MD) method. The proposed method not only inherits the simplicity of the widely used MD method, but also takes the lags between different spatial points into account when constructed the cross-correlation matrix. In addition, by determining a non-correlation distance, the size of correlation range and dimension of the correlation matrix are significantly reduced, which solves the problem of insufficient computer memory and considerably improves the efficiency of random field inversion. Finally, taking the soil shear strength and stiffness parameters as a case in point, a multivariate three-dimensional cross-correlated random field is established to illustrate the effectiveness of M-MD method. The M-MD method facilitates three-dimensional spatial variability modeling in probabilistic site characterization and routine geotechnical reliability analysis. |
| 学科主题 | Science & Technology - Other Topics |
| 语种 | 英语 |
| WOS记录号 | WOS:000991649800002 |
| 出版者 | SPRINGER HEIDELBERG |
| 源URL | [http://119.78.100.198/handle/2S6PX9GI/35470]  |
| 专题 | 中科院武汉岩土力学所 |
| 作者单位 | 1.Chinese Academy of Sciences; Ningbo University of Technology 2.Chinese Academy of Sciences; University of Chinese Academy of Sciences, CAS; 3.Chinese Academy of Sciences; Wuhan Institute of Rock & Soil Mechanics, CAS; |
| 推荐引用方式 GB/T 7714 | Tian, Ning,Chen, Jian,Wang, Zhansheng,et al. Modified Matrix Decomposition Method for Generating Multivariate Three-Dimensional Cross-Correlated Random Fields[J]. ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING,2023. |
| APA | Tian, Ning,Chen, Jian,Wang, Zhansheng,Tian, Kaiwei,&Huang, Juehao.(2023).Modified Matrix Decomposition Method for Generating Multivariate Three-Dimensional Cross-Correlated Random Fields.ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING. |
| MLA | Tian, Ning,et al."Modified Matrix Decomposition Method for Generating Multivariate Three-Dimensional Cross-Correlated Random Fields".ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING (2023). |
入库方式： OAI收割
来源：武汉岩土力学研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。