近年来，人工智能在各个领域展现出了强大的模式识别和分类能力，为岩性识别提供了新的思路。从支持向量机、神经网络、集成学习这3种方法出发，综述这些机器学习算法的基本原理、优缺点及其在铀矿层岩性识别领域的研究进展和应用情况。结果表明：机器学习通过训练模型可以有效识别出测井数据与不同岩性之间的关联，将岩性识别过程转化为机器学习的过程，可以极大地提高岩性识别自动化程度和识别准确率，具有重要的现实意义和广阔的发展前景。

In recent years, artificial intelligence has demonstrated strong pattern recognition and classification capabilities across various fields, providing new insights for lithology identification. Starting from three methods: support vector machines, neural networks, and ensemble learning, the basic principles, advantages and disadvantages of these machine learning algorithms were reviewed, as well as their research progress and application in the field of uranium ore bed lithology identification. The results show that machine learning can effectively identify the correlation between logging data and different lithologies through model training, transforming the process of lithology identification into a machine learning process. This can greatly improve the automation level and accuracy of lithology identification, holding significant practical importance and a broad development prospect.