Lithological Facies Classification Using Attention-Based Gated Recurrent Unit

• DOI: 10.26599/tst.2023.9010077
• Publication Date: 2024-02-09T18:35:21Z
• Authors: Yuwen Liu; Yulan Zhang; Xingyuan Mao; Xucheng Zhou; Jingwen Chang; Wenwei Wang; Pan Wang; Lianyong Qi

Lithological facies classification is a pivotal task in petroleum geology, underpinning reservoir characterization and influencing decision-making exploration production operations. Traditional methods, such as support vector machines Gaussian process classifiers, often struggle with the complexity nonlinearity of geological data, leading to suboptimal performance. Moreover, numerous prevalent approaches fail adequately consider inherent dependencies sequence measurements from adjacent depths well. A novel approach leveraging an attention-based gated recurrent unit (AGRU) model introduced this paper address these challenges. The AGRU excels by exploiting sequential nature well-log data capturing long-range through attention mechanism. This enables flexible context-dependent weighting different parts sequence, enhancing discernment key features for classification. proposed method was validated on two publicly available datasets. Results demonstrate considerably improvement over traditional methods. Specifically, achieved superior performance metrics considering precision, recall, F1-score.