基于钻进参数聚类的含煤地层岩性模糊识别

通过钻进参数进行煤矿巷道围岩特征描述可为煤矿安全绿色开采提供地质信息保障。针对煤矿井下坑道钻探中随钻地层岩性识别难度大、精度低的问题,提出了一种基于钻进参数核模糊C均值聚类(Kernel Fuzzy C-means,KFCM)算法的含煤地层岩性模糊识别方法。结合钻进试验台上开展的模拟岩样钻进试验,获得了包括钻速、转速和钻压等敏感钻进参数的训练样本,利用KFCM算法对获取的钻进参数训练样本进行学习,构造钻进参数样本空间并映射到高维空间进行聚类处理。建立了以典型含煤地层分类为目标的聚类模型,采用高斯核函数分别确定了软弱夹层、煤层和泥岩层的分布结构以及对应的聚类中心。其中,对比线性核函数,高斯核函数在垂向上的分类效果符合沉积岩构造的特征,且聚类时间节约了7.2%。进一步基于钻进参数的聚类结果,将钻速作为衡量各类岩石钻进性能的关键参数,通过分析钻进参数数据集的变化规律,建立了钻速与转速、钻压幂函数表达形式的地层岩性预测模型,采用数据插值拟合方法完成了典型软弱夹层、煤层和泥岩层的空间划分。并应用模糊数学方法通过构建钻速的分段三角形隶属度函数,得出样本地层钻速对典型含煤地层钻速的隶属度公式,根据隶属度公式将地层岩性划分为5个级别,实现了对样本地层岩性的模糊识别。在实钻试验中,对提出的模糊识别方法的有效性进行了验证。结果表明,该方法能够在PDC锚杆钻头回转钻进条件下快速识别典型含煤地层岩性,识别的正确率为92%,研究结果为实现煤矿井下巷道隐蔽致灾因素动态智能探测提供了借鉴。

Fuzzy identification of coal bearing strata lithology based on drilling parameter clustering

Evaluation of surrounding rock characteristics using drilling parameters can provide support for safe and green coal mining.Due to the significant difficulty and low accuracy in lithology identification during exploration in underground coal mine roadway,a fuzzy identification method for coal bearing strata based on the Kernel Fuzzy C means(KFCM) clustering of drilling parameters is proposed.In combination with simulated rock sample drilling tests conducted on the drilling bench,some training samples aimed at understanding sensitive drilling parameters such as drilling rate,rotation speed,and bit pressure are obtained.The KFCM algorithm is used to analyze drilling parameter training samples,and a drilling parameter sample space is constructed and mapped to high dimensional space for clustering.The distribution structures in the weak interlayer,coal seam,and mudstone layer and the corresponding clustering centers are determined using a Gaussian kernel function.Compared to the linear kernel function,the vertical classification of the Gaussian kernel function better aligns with the characteristics of sedimentary rock structure,and clustering time is reduced by 7.2%.Based on the clustering results of the drilling parameters,the drilling rate is taken as the key parameter to measure the drilling performance of multiple rock types.By analyzing variation in the drilling parameter data set,a prediction model for drilling rate with a power function of bit pressure and rotation speed is established.The spatial division of typical weak interlayer coal seam and mudstone layer is completed using data interpolation.The membership formula of the drilling rate in a typical coal bearing formation is obtained by constructing the subsection triangle membership function of drilling rate using the fuzzy mathematical method.According to the membership formula,the lithology is divided into five grades,which help to realize the fuzzy identification different lithologies.The validity of the proposed method is verified using a practical drilling experiment.Results show that this method can quickly identify the lithology of typical coal bearing strata under a rotary PDC anchor bit,with an accuracy of 92%.The preset research provides an effective method for identifying potential disaster causing factors in coal mine roadway.