Research status and prospects of intelligent logging lithology identification

Abstract With the increasing of petroleum exploration and development, accurate lithology identification is of crucial. Machine learning (ML) plays a key role in logging lithology identification. By introducing traditional logging lithology identification methods, we review the application of ML in logging lithology identification from the perspectives of bibliometrics and ML classification in this paper. The applications of supervised learning, semi-supervised learning, unsupervised learning, ensemble learning, and deep learning algorithms in logging lithology identification are introduced in detail. Multiple ML algorithms have achieved remarkable results in different scenarios. For example, support vector machine, random forest, eXtreme gradient boosting, and convolutional neural network perform well in logging lithology identification and obtain relatively high identification accuracy. However, ML for logging lithology identification also faces challenges such as data quality, data imbalance, model generalization, and model interpretability. Future research should focus on algorithm optimization and innovation, improvements in data quality and quantity, deep multidisciplinary integration and practical application to enhance the accuracy and reliability of lithology identification. These findings provide strong support for oil and gas exploration and development.