Application of machine learning for facies recognition in An Chau basin

Abstract

The An Chau basin (An Chau basin) is a geological structure extending in a northwest-southeast direction, located in the northeastern region of Vietnam. The An Chau basin is considered a large-scale basin with significant petroleum potential. This fact suggests a high likelihood of discovering hydrocarbon accumulations within the An Chau basin within Vietnam’s territory. Although the petroleum potential of this basin was identified early, for various reasons, geological surveys and exploration activities in the An Chau basin remain rudimentary and have not yet met the requirements for effective petroleum exploration and production. The application of machine learning (ML) models to lithofacies identification
offers a novel approach to reducing processing time, consolidating large and diverse datasets, and uncovering hidden relationships among layers of identification information. The primary objective of this study is to identify lithofacies in the An Chau basin using comprehensive datasets trained with a decision tree (DT) structure combined with a gradient boosting algorithm (XGB) to evaluate the structure and assess the petroleum potential of this area. A prerequisite for improving the accuracy of machine learning is enriching the database through the integration of geological and seismic data, along with the calculation of additional attributes to build the “learning model” training the ML system and applying the trained model to identify lithofacies in the An Chau basin.