Home

Quick Links

Search

 
Research on Adaptive Petrophysical Modeling Based on Machine Learning and Multivariate GeostatisticsNormal access

Authors: B. Yang and Z. Liu
Event name: 81st EAGE Conference and Exhibition 2019
Session: Potential Fields Image Analysis Tools and ML Advances
Publication date: 03 June 2019
DOI: 10.3997/2214-4609.201900966
Organisations: EAGE
Language: English
Info: Extended abstract, PDF ( 902.93Kb )
Price: € 20

Summary:
3D gravity inversion plays an important l role in the quantitative interpretation of practical gravity data. One of the key issues with 3D inversion of gravity data is the multiplicity. Combining multiple geophysical data is an advantageous means of reducing multiplicity. However, establishing a petrophysical relationship between different physical data is a major difficulty. We propose a process for petrophysical modeling using machine learning and multiple geostatistics.Based on the Fuzzy c-means (FCM) and an adaptive cross- variogram function fitting(which make it possible to introduce the cross-variogram in multivariate geostatistics into the traditional objective function), we can better suggest the spatial correlation of petrophysical. Synthetic example demonstrated the feasibility and reliability of our method.


Back to the article list