State Key Laboratory of the Gas Disaster Detecting, Preventing and Emergency Controlling
CCTEG Chongqing Research Institute

针对底抽巷瓦斯抽采穿层钻孔施工过程中，煤岩界面识别不及时、不准确，缺少相应技术手段的问题，设计开发了一套基于钻进参数(转速、回转扭矩、推进力、推进速度、破碎比功)的煤岩界面识别系统，整套系统由数据感知层、采集层和分析层组成。其中，数据感知层和数据采集层合称钻机数据采集系统，可以对钻进参数进行实时采集；数据分析层则采用支持向量机(Support Vector Machine, SVM)分类算法对带有煤岩分类标记的钻进参数进行数据学习和模型训练，继而对未知的钻进参数进行分类预测，最终实现煤岩界面自动识别。在河南鹤壁中泰矿业的现场应用表明：钻进参数中的回转扭矩、推进速度和破碎比功在煤岩界面处均产生明显的“涨落”，可以作为区分煤层和岩石的3个特征参数；使用线性核函数的支持向量机分类模型可以准确地将两种地层中的钻进参数区分出来，通过对训练集中89个样本数据学习即可在测试集中获得100%的正确率，说明了特征参数和地层信息之间是线性可分的。该系统推广应用不仅可以为煤岩分类识别提供基础数据的获取途径；还可以为穿层钻孔的煤岩界面识别提供一定的科学依据和指导，从而确保钻孔达标，避免抽采空白带的产生。

In response to the problems of untimely and inaccurate coal-rock interface recognition and lack of appropriate technical means during the construction of cross-seam drilling for gas drainage by bottom drainage roadway, a coal-rock interface recognition system based on drilling parameters (rotational speed, rotary torque, propulsion force, advance velocity, crushing work ratio) was developed. The entire system consists of a data sensing layer, a data acquisition layer and a data analysis layer. Among them, the data sensing layer and the data acquisition layer are also collectively called the drilling data acquisition system, which can collect the drilling parameters in real time. The data analysis layer performs the data learning and model training for the drilling parameters with coal or rock classification labels using the Support Vector Machine (SVM) classification algorithm, then classifies and predicts the unknown drilling parameters, and ultimately achieves the automatic recognition of coal-rock interface. The field application of Zhongtai Mining in Hebi, Henan shows that: the rotary torque, advance velocity and crushing work ratio fluctuate significantly at the coal-rock interface, and thus they can be regarded as the three characteristic parameters to distinguish the coal and rock. The SVM classification model using linear kernel functions can accurately distinguish the drilling parameters in the two types of formations. By learning from the 89 sample data in the training set, a 100% accuracy rate can be obtained in the test set, which also indicates that the characteristic parameters and the formation information are linearly separable. Generally, the promotion and application of this system can not only provide a way to obtain the basic data for coal-rock classification and identification, but also provide certain scientific basis and guidance for the identification of coal-rock interface recognition in cross-seam drilling, thereby ensuring the standardized drilling and avoiding the occurrence of unproductive zones.

天地科技股份有限公司科技创新创业资金专项项目(2022-2-TD-ZD008)；中煤科工集团重庆研究院有限公司创新引导项目(2021YBXM49)

王杰. 基于钻进参数的煤岩界面识别系统研究[J]. 煤田地质与勘探，2023，51(9)：72−80.

WANG Jie. Research on coal-rock interface recognition system based on drilling parameters[J]. Coal Geology & Exploration，2023，51(9)：72−80.