New variogram modeling method using MGGP and SVR |
| |
| Authors: | Changik Han Jiyang Wang Mingguo Zheng Ende Wang Jianming Xia GwangSu Li Sunchol Choe |
| |
| Institution: | 1.College of Resources & Civil,Northeastern University,Shenyang,China;2.College of Geoexploration Engineering,Kimchaek University of Technology,Pyongyang,Democratic People’s Republic of Korea;3.College of Information Science & Engineering,Northeastern University,Shenyang,China |
| |
| Abstract: | A critical step for kriging in geostatistics is estimation of the variogram. Traditional variogram modeling comprise of the experimental variogram calculation, appropriate variogram model selection and model parameter determination. Selecting of the variogram model and fitting of model parameters is the most controversial aspect of geostatistics. Shapes of valid variogram models are finite, and sometimes, the optimal shape of the model can not be fitted, leading to reduced estimation accuracy. In this paper, a new method is presented to automatically construct a model shape and fit model parameters to experimental variograms using Support Vector Regression (SVR) and Multi-Gene Genetic Programming (MGGP). The proposed method does not require the selection of a variogram model and can directly provide the model shape and parameters of the optimal variogram. The validity of the proposed method is demonstrated in a number of cases. |
| |
| Keywords: | |
| 本文献已被 SpringerLink 等数据库收录！ |
|
