地层岩体种类的识别是地质和岩土勘察工作的重点内容，为解决地层界面岩性辨识困难的难题，研发了一种随钻测量系统，提出了相应的数据处理及分析方法。采用基于机器学习技术的方法，选取支持向量机（SVM）算法，研究了一种通过随钻参数反演地层岩体种类的方案。得出了如下结论： 建立各参数同孔深数据的对应关系，有利于机器学习数据库的建立和反演方案的实施。随钻监测数据处理原则是: 识别各种非钻进状态及异常状态，建立纯钻进状态的时间—孔深曲线和各随钻参数—孔深变化的曲线。SVM在地层岩体种类识别方面取得了良好的效果，预测结果与钻探记录基本一致，可用于地层关键岩体种类识别，防止虚假钻孔和虚假编录，为岩土工程智能勘测提供了新的途径。

The identification of stratigraphic lithology is a key element of geological and geotechnical surveys. In this paper, a drilling measurement system is developed, and the corresponding data processing and analysis methods are proposed. Based on machine learning technology, a support vector machine (SVM) algorithm is selected to examine a scheme to invert the formation lithology by the drilling parameters. It is shown that establishing the correspondence between each parameter and borehole depth data is beneficial to the establishment of machine learning database and the implementation of the inversion scheme. The drilling measurement parameters processing principle is identifying the stop and abnormal status, and establishing drilling timehole depth and drilling parameterhole depth curves. SVM achieves good results in formation lithology identification, and the prediction results are basically consistent with drilling records, which can be used for key formation lithology identification, preventing false drilling and false compilation, and providing a new way for intelligent surveying in geotechnical engineering.