Coal Geology & Exploration
Abstract
Objective Vibration signals while drilling reflect the physical properties and microstructural characteristics of rocks, providing a valuable basis for real-time lithology identification. However, existing models suffer from limited accuracy and robustness due to the low dimensionality of signal features and the class imbalance induced by variations in the rock layer thickness. Methods To overcome these limitations, this study proposed a novel lithology identification method based on a bidirectional long short-term memory (BiLSTM) network that integrates attention mechanism (also referred to as the BiLSTM-Attention). This approach incorporates automated feature extraction and class weight adjustment strategies to enhance both the classification performance and generalizability of the model. First, triaxial vibration acceleration signals were acquired using a custom-designed measurement sub. Subsequently, a wide range of statistical, time-domain, and frequency-domain features were automatically extracted utilizing Tsfresh—a Python package. Key features were then selected through feature importance evaluation, followed by the construction of a high-quality feature space. Afterward, temporal dependencies inherent in the time-series data were modeled using the BiLSTM network. Meanwhile, attention mechanism was introduced to enhance critical features through weighting, thus improving the sensitivity to lithological variations of the resulting model. In addition, a class weight calculation mechanism was adopted to reduce the adverse impact of class imbalance, thereby enhancing the ability of the model to accurately identify a minority of lithologies. Results and Conclusions The experimental results indicate that for the dataset treated with feature engineering using Tsfresh, the proposed BiLSTM-Attention model yielded a training accuracy of 99.59% and a low loss value of 0.0125, which increased by 12.35% and decreased by 0.3915, respectively compared to raw data without feature engineering. The proposed model yielded precision ranging from 98.31% to 100.00%, recall from 98.99% to 100.00%, and F1 scores from 99.15% to 99.73%, increasing by 4.23% to 6.89%, 4.46% to 5.13%, and 4.7% to 5.38%, respectively compared to those of the GRU, LSTM, BiLSTM, and LSTM-Attention models. Furthermore, the proposed method outperformed traditional models in terms of feature representation and classification accuracy. Notably, it exhibited enhanced robustness and stability in identifying a minority of lithologies. Overall, this study provides a novel approach for intelligent lithology identification while drilling.
Keywords
lithology identification, vibration signal, feature extraction, bidirectional long short-term memory (BiLSTM), attention mechanism
DOI
10.12363/issn.1001-1986.25.04.0262
Recommended Citation
WANG Xi, PENG Fengjia, LI Siyang,
et al.
(2025)
"Intelligent lithology identification while drilling using BiLSTM integrating attention mechanism,"
Coal Geology & Exploration: Vol. 53:
Iss.
11, Article 22.
DOI: 10.12363/issn.1001-1986.25.04.0262
Available at:
https://cge.researchcommons.org/journal/vol53/iss11/22
Reference
[1] 曹宏，常德双，胡少华，等. 中国深层油气勘探地球物理技术与展望[J]. 石油学报，2023，44(12)：2250−2269.
CAO Hong，CHANG Deshuang，HU Shaohua，et al. Geophysical techniques and prospects for deep oil–gas exploration in China[J]. Acta Petrolei Sinica，2023，44(12)：2250−2269.
[2] GAMAL H，ELKATATNY S，ABDULAZIZ A M. Intelligent solution for auto–detecting lithology scheme while drilling by machine learning[C]//2024 International Petroleum Technology Conference. Dhahran：IPTC，2024：D031S095R004.
[3] OLORUNTOBI O，BUTT S. Application of specific energy for lithology identification[J]. Journal of Petroleum Science and Engineering，2020，184：106402.
[4] 王胜，赖昆，张拯，等. 基于随钻振动信号与深度学习的岩性智能预测方法[J]. 煤田地质与勘探，2023，51(9)：51−63.
WANG Sheng，LAI Kun，ZHANG Zheng，et al. Intelligent lithology prediction method based on vibration signal while drilling and deep learning[J]. Coal Geology & Exploration，2023，51(9)：51−63.
[5] 梁栋才，汤华，吴振君，等. 基于多钻进参数和概率分类方法的地层识别研究[J]. 岩土力学，2022，43(4)：1123−1134.
LIANG Dongcai，TANG Hua，WU Zhenjun，et al. Stratum identification based on multiple drilling parameters and probability classification[J]. Rock and Soil Mechanics，2022，43(4)：1123−1134.
[6] 陈见行，曾班全，张俊文. 加卸载条件下冲击性岩石动态力学特性研究[J]. 岩石力学与工程学报，2024，43(10)：2430−2442.
CHEN Jianhang，ZENG Banquan，ZHANG Junwen. Dynamic mechanical performance of impact rocks under loading and unloading condition[J]. Chinese Journal of Rock Mechanics and Engineering，2024，43(10)：2430−2442.
[7] ALZUBAIDI F，MOSTAGHIMI P，SWIETOJANSKI P，et al. Automated lithology classification from drill core images using convolutional neural networks[J]. Journal of Petroleum Science and Engineering，2021，197：107933.
[8] 罗仁泽，庹娟娟，倪华玲，等. 基于改进集成学习的测井岩性识别方法研究[J]. 石油物探，2023，62(2)：212−224.
LUO Renze，TUO Juanjuan，NI Hualing，et al. Logging lithology identification method based on improved ensemble learning[J]. Geophysical Prospecting for Petroleum，2023，62(2)：212−224.
[9] CARPENTER C. Machine–learning techniques classify，quantify cuttings lithology[J]. Journal of Petroleum Technology，2024，76(1)：92−94.
[10] 文士豪，李忠诚，田彦玲，等. 四川盆地南部地区茅口组碳酸盐岩储层有效性评价及流体识别新方法[J]. 天然气工业，2024，44(12)：15−26.
WEN Shihao，LI Zhongcheng，TIAN Yanling，et al. Reservoir effectiveness evaluation and new fluid identification method of Maokou Formation carbonate rocks in southern Sichuan Basin[J]. Natural Gas Industry，2024，44(12)：15−26.
[11] 朱怡翔，石广仁. 火山岩岩性的支持向量机识别[J]. 石油学报，2013，34(2)：312−322.
ZHU Yixiang，SHI Guangren. Identification of lithologic characteristics of volcanic rocks by support vector machine[J]. Acta Petrolei Sinica，2013，34(2)：312−322.
[12] 徐鹏宇，周怀来，赵霞，等. 基于极限学习机的碳酸盐岩储层测井评价方法：以川中北部GM区块灯二段为例[J]. 大庆石油地质与开发，2022，41(6)：133−142.
XU Pengyu，ZHOU Huailai，ZHAO Xia，et al. Logging evaluation method for carbonate reservoir based on extreme learning machine：A case study of Member 2 of Dengying Formation in GM block in north central Sichuan Basin[J]. Petroleum Geology & Oilfield Development in Daqing，2022，41(6)：133−142.
[13] ZOU Yanhong，CHEN Yuting，DENG Hao. Gradient boosting decision tree for lithology identification with well logs：A case study of Zhaoxian gold deposit，Shandong peninsula，China[J]. Natural Resources Research，2021，30(5)：3197−3217.
[14] HAN Ruiyi，WANG Zhuwen，WANG Wenhua，et al. Lithology identification of igneous rocks based on XGboost and conventional logging curves，a case study of the eastern depression of Liaohe Basin[J]. Journal of Applied Geophysics，2021，195：104480.
[15] 李瑞，吴文圣. 基于SSA–XGBoost模型的高精度密度测井预测方法研究[J]. 核技术，2024，47(12)：120502.
LI Rui，WU Wensheng. SSA–XGBoost model based high–precision density prediction method for well logging[J]. Nuclear Techniques，2024，47(12)：120502.
[16] 王新领，祝新益，张宏兵，等. 基于随机树嵌入的随钻测井岩性识别方法[J]. 吉林大学学报(地球科学版)，2024，54(2)：701−708.
WANG Xinling，ZHU Xinyi，ZHANG Hongbing，et al. Lithology identification method for logging while drilling based on random tree embedding[J]. Journal of Jilin University (Earth Science Edition)，2024，54(2)：701−708.
[17] 蔡明，周庆文，杨聪，等. 砂泥岩地层岩性智能识别方法与应用：以新疆轮南侏罗系岩层为例[J]. 煤田地质与勘探，2025，53(1)：235−244.
CAI Ming，ZHOU Qingwen，YANG Cong，et al. An intelligent lithology identification method for sandstone and mudstone strata and its applications：A case study of the Jurassic strata in the Lunnan area，Xinjiang，China[J]. Coal Geology & Exploration，2025，53(1)：235−244.
[18] ASHRAF U，ZHANG Hucai，ANEES A，et al. A core logging，machine learning and geostatistical modeling interactive approach for subsurface imaging of lenticular geobodies in a clastic depositional system，SE Pakistan[J]. Natural Resources Research，2021，30(3)：2807−2830.
[19] 熊冬，王翔，马新仿，等. 基于机械比能聚类分析的深层煤岩水平段岩性判识方法[J]. 天然气工业，2025，45(2)：114−124.
XIONG Dong，WANG Xiang，MA Xinfang，et al. Lithology identification of deep coal horizontal section based on mechanical specific energy cluster analysis[J]. Natural Gas Industry，2025，45(2)：114−124.
[20] 贾泽庆，吴振君，汤华，等. 冲击回转钻孔过程振动响应时频特征与岩石力学性质相关性研究[J]. 振动与冲击，2024，43(21)：245−252.
JIA Zeqing，WU Zhenjun，TANG Hua，et al. Correlation between time–frequency characteristics of vibration response in percussion–rotary drilling process and rock mechanical properties[J]. Journal of Vibration and Shock，2024，43(21)：245−252.
[21] 杨巍,田帅康,郭大超,等. 潜孔钻机随钻参数对钻孔效率的影响研究[J]. 地质与勘探,2025,61(1):154−161
YANG Wei,TIAN Shuaikang,GUO Dachao,et al. The influence of drilling parameters of downhole drilling rig on drilling efficiency[J]. Geology and Exploration,2025,61(1):154−161
[22] 雷志勇，王家文，凡东，等. 基于1DCNN–BiLSTM–CBAM的煤巷顶板岩层地质特征随钻智能识别方法[J]. 煤田地质与勘探，2024，52(11)：192−199.
LEI Zhiyong，WANG Jiawen，FAN Dong，et al. An intelligent identifying–while–drilling method for geological features of roof strata in coal roadways based on a 1DCNN–BiLSTM–CBAM model[J]. Coal Geology & Exploration，2024，52(11)：192−199.
[23] 刘金锁，张平松，刘盛东，等. 基于EEMD和DE–PNN的随钻测震岩性识别技术研究[J/OL]. 煤炭学报，2025：1–12 [2025-03-24]. https://link.cnki.net/doi/10.13225/j.cnki.jccs.KT24.1573
LIU Jinsuo，ZHANG Pingsong，LIU Shengdong，et al. Research on lithology identification technology of seismic measurement while drilling based on EEMD and DE–PNN[J/OL]. Journal of China Coal Society，2025：1–12 [2025-03-24]. https://link.cnki.net/doi/10.13225/j.cnki.jccs.KT24.1573.
[24] FLEGNER P，KAČUR J，DURDÁN M，et al. Processing a measured vibroacoustic signal for rock type recognition in rotary drilling technology[J]. Measurement，2019，134：451−467.
[25] LAZAROVÁ E，KRUĽÁKOVÁ M，KRÚPA V，et al. Regime and rock identification in disintegration by drilling based on vibration signal differentiation[J]. International Journal of Rock Mechanics and Mining Sciences，2022，149：104984.
[26] CHRIST M，BRAUN N，NEUFFER J，et al. Time series feature extraction on basis of scalable hypothesis tests (tsfresh–A Python package)[J]. Neurocomputing，2018，307：72−77.
[27] 张欣，宋雨鑫. 人工智能准确性原则的理论阐释与制度展开[J]. 电子政务，2025(3)：110–121
[28] 王桡，鄢社锋，毛琳琳，等. 基于通道分组注意力机制的水声目标识别网络[J]. 信号处理，2025，41(3)：524−532.
WANG Rao，YAN Shefeng，MAO Linlin，et al. Underwater acoustic target recognition network based on channel grouping attention mechanism[J]. Journal of Signal Processing，2025，41(3)：524−532.
[29] RAZALI N M，YAP B W. Power comparisons of Shapiro–Wilk，Kolmogorov–Smirnov，Lilliefors and Anderson–darling tests[J]. Journal of Statistical Modeling and Analytics，2011，2(1)：21−33.
[30] BENJAMINI Y，YEKUTIELI D. The control of the false discovery rate in multiple testing under dependency[J]. The Annals of Statistics，2001，29(4)：1013699998.
[31] SIAMI–NAMINI S，TAVAKOLI N，NAMIN A S. The performance of LSTM and BiLSTM in forecasting time series[C]//2019 IEEE International Conference on Big Data (Big Data). Los Angeles：IEEE，2019：3285–3292.
[32] NIU Zhaoyang，ZHONG Guoqiang，YU Hui. A review on the attention mechanism of deep learning[J]. Neurocomputing，2021，452：48−62.
[33] ZHANG Cong，GAO Ke，LYU Xiaoshu，et al. Experimental study on the rock–breaking mechanism of annular–grooved structure and rotary direction for impregnated diamond bits[J]. Geoenergy Science and Engineering，2025，249：213812.
[34] MAHARANA K，MONDAL S，NEMADE B. A review：Data pre–processing and data augmentation techniques[J]. Global Transitions Proceedings，2022，3(1)：91−99.
[35] HEYDARIAN M，DOYLE T E，SAMAVI R. MLCM：Multi–label confusion matrix[J]. IEEE Access，2022，10：19083−19095.
Included in
Earth Sciences Commons, Mining Engineering Commons, Oil, Gas, and Energy Commons, Sustainability Commons