基于PCA-PSO-SVM的煤岩可钻性预测方法  

Coal-rock Drillability Prediction Method Based on PCA-PSO-SVM

作　　者：甘林堂[1] 张幼振 张磊[1] 陈韬[3] 张凯 姚克[2] 宋海涛 GAN Lintang;ZHANG Youzhen;ZHANG Lei;CHEN Tao;ZHANG Kai;YAO Ke;SONG Haitao(Huainan Mining(Group)Co.,Ltd.,Huainan,Anhui 232001;China Coal Technology&Engineering Xi’an Research Institute(Group)Co.,Ltd.,Xi’an,Shaanxi 710077;Chinese Institute of Coal Science,Beijing 100013)

机构地区：[1]淮南矿业(集团)有限责任公司,安徽淮南232001 [2]中煤科工西安研究院(集团)有限公司,陕西西安710077 [3]煤炭科学研究总院,北京100013

出　　处：《中国煤炭地质》2025年第3期40-44,共5页Coal Geology of China

基　　金：陕西省自然科学基础研究重点项目(2024JC-ZDXM-30);陕西省重点研发计划项目(2023-YBGY-340)。

摘　　要：煤岩可钻性的预测是实现煤矿井下智能化钻探的基础。提出一种以钻进参数作为可钻性指标的分级方法,从钻进参数中选取4种影响岩石可钻性的等级因素,用主成分分析法(PCA)解释每种影响因素之间的相关性及贡献率,降低数据维度的同时提高预测能力。通过粒子群优化和支持向量机(PSO-SVM)算法开发,合理设置预测模型参数值。以淮南矿区现场实钻数据作为样本基础,建立煤岩可钻性预测模型。通过优化前后机器学习算法模型的预测对比结果表明,提出的预测方法对煤岩可钻性等级预测准确率达到97.5%,预测准确率相比传统方法更高。研究结果可以为煤矿井下钻进过程中的地层识别,实时优化钻机操控参数,实现自适应钻进控制提供理论依据。The prediction of coal rock drillability is the foundation for achieving intelligent drilling in coal mines underground.Propose a grading method based on drilling parameters as drillability indicators,selecting four grading factors that affect rock drillability from drilling parameters,and using principal component analysis(PCA)to explain the correlation and contribution rate between each influencing factor,reducing data dimensionality while improving predictive ability.By using particle swarm optimization and support vector machine(PSO-SVM)algorithm development,the parameter values of the prediction model are reasonably set.Establish a coal-rock drillability prediction model based on on-site drilling data from Huainan mining area as the sample basis.The comparison results of machine learning algorithm models before and after optimization show that the proposed prediction method achieves an accuracy of 97.5%in predicting the drillability level of coal and rock,which is higher than traditional methods.The research results can provide theoretical basis for the identification of strata during coal mine underground drilling,real-time optimization of drilling rig control parameters,and realization of adaptive drilling control.

关 键 词：煤岩可钻性 主成分分析法 PSO-SVM算法 钻进参数 预测模型 淮南矿区 

分 类 号：P634[天文地球—地质矿产勘探]