BP Neural Network-Based Models to Predict Clay Minerals

doi:10.7657/XJPG20210517

Abstract

Abstract:

Accurate prediction of clay minerals is the key to deep drilling operation and pay zone protection. In order to determine the distribution law of clay minerals in the Jurassic Ahe formation in the northern tectonic zone of the Kuqa depression, Tarim basin, a well logging model and a combined model based on BP neural network were constructed by using GR logging parameters, cation exchange capacity, hydrogen index and photoelectric absorption cross-section index. The average absolute errors of the two models are 5.34% and 2.38%, respectively. Applied to Well Yinan 5, the average absolute errors of the models are 4.64% and 3.45%, respectively, by considering X-ray diffraction data. The prediction result shows that the clay mineral contents from high to low are illite, chlorite, illite/smectite mixed layer and kaolinite in Well Yinan 5. Damages from velocity sensitivity and acid sensitivity should be prevented in reservoir development.

Key words: clay mineral, GR logging, BP neural network, cation exchange capacity, hydrogen index, photoelectric absorption cross-section index

CLC Number:

TE112

LI Xinyu, OUYANG Chuanxiang, YANG Bowen, ZHAO Hongnan, NIE Bin. BP Neural Network-Based Models to Predict Clay Minerals[J]. Xinjiang Petroleum Geology, 2021, 42(5): 624-629.

Figures/Tables 9

Table 1

Table 2

Table 3

Fig. 1.

Fig. 2.

Table 4

Fig. 3.

Table 5

Table 6

References 15

[1]	刘菁华, 王祝文, 易清平. 利用自然γ能谱测井资料确定粘土矿物的含量及其应用[J]. 吉林大学学报(地球科学版), 2010, 40(1):215-221.
	LIU Jinghua, WANG Zhuwen, YI Qingping. Determinating clay mineral content by natural gamma-ray spectral logging data[J]. Journal of Jilin University (Earth Science Edition), 2010, 40(1):215-221.
[2]	杨大超. 自然伽马能谱曲线在大牛地气田粘土岩类型分析中的应用[J]. 石油天然气学报, 2007, 29(3):400-402.
	YANG Dachao. Application of natural gamma ray spectrum curve in clay rock type analysis of Daniudi gas field[J]. Journal of Oil and Gas Technology, 2007, 29(3):400-402.
[3]	宁波, 姚志刚, 李瑞祥. 自然伽马能谱测井在延安地区黏土矿物分析中的应用[J]. 科学技术与工程, 2020, 20(9):3 435-3 440.
	NING Bo, YAO Zhigang, LI Ruixiang. Application of natural gamma spectrometric logging for clay mineral analysis in Yan’an area[J]. Science Technology and Engineering, 2020, 20(9):3 435-3 440.
[4]	于聪灵, 蔡忠贤, 杨海军, 等. 基于BP神经网络预测轮古油田奥陶系碳酸盐岩油藏洞穴充填程度[J]. 新疆石油地质, 2018, 39(5):614-621.
	YU Congling, CAI Zhongxian, YANG Haijun, et al. Prediction of cave filling degree in Ordovician carbonate reservoirs based on BP neural network in Lungu oilfield,Tarim basin[J]. Xinjiang Petroleum Geology, 2018, 39(5):614-621.
[5]	冯渊, 秦康, 李二党, 等. 基于黏土矿物分析的低电阻率油层识别[J]. 新疆石油地质, 2020, 41(2):237-242.
	FENG Yuan, QIN Kang, LI Erdang, et al. Identification of low-resistivity oil layers based on clay mineral analysis[J]. Xinjiang Petroleum Geology, 2020, 41(2):237-242.
[6]	黎盼, 孙卫, 李长政. 鄂尔多斯盆地华庆油田长6₃储集层成岩相特征[J]. 新疆石油地质, 2018, 39(5):517-523.
	LI Pan, SUN Wei, LI Changzheng. Diagenetic facies characteristics of Chang 6₃ reservoir in Huaqing oilfield,Ordos basin[J]. Xinjiang Petroleum Geology, 2018, 39(5):517-523.
[7]	尤源, 梁晓伟, 冯胜斌, 等. 鄂尔多斯盆地长7段致密储层主要黏土矿物特征及其地质意义[J]. 天然气地球科学, 2019, 30(8):1 233-1 241.
	YOU Yuan, LIANG Xiaowei, FENG Shengbin, et al. Features and geological significance of main clay minerals in Chang 7 tight sandstone reservoir,Ordos basin[J]. Natural Gas Geoscience, 2019, 30(8):1 233-1 241.
[8]	FERTLW H, CHILINGARIANG V. Type and distribution modes of clay minerals from well logging data[J]. Journal of Petroleum Science ＆ Engineering, 1990, 3(4):321-332.
[9]	邢培俊, 孙建孟, 王克文, 等. 利用测井资料确定粘土矿物的方法对比[J]. 中国石油大学学报(自然科学版), 2008, 32(2):53-57.
	XING Peijun, SUN Jianmeng, WANG Kewen, et al. Comparison of determination methods for clay minerals using log data[J]. Journal of China University of Petroleum(Edition of Natural Science), 2008, 32(2):53-57.
[10]	张鹏, 吴通, 李中, 等. BP神经网络法预测顺北超深碳酸盐岩储层应力敏感程度[J]. 石油钻采工艺, 2020, 42(5):622-626.
	ZHANG Peng, WU Tong, LI Zhong, et al. Application of BP neural network method to predict the stress sensitivity of ultra deep carbonate reservoir in Shunbei oilfield[J]. Oil Drilling & Production Technology, 2020, 42(5):622-626.
[11]	赵冰. 利用核测井资料预测海相页岩储集层总有机碳含量[J]. 新疆石油地质, 2019, 40(4):499-504.
	ZHAO Bing. Predicting total organic carbon content in marine shale reservoirs with nuclear logging data[J]. Xinjiang Petroleum Geology, 2019, 40(4):499-504.
[12]	郑欣, 阴国锋, 王健伟, 等. 东海盆地西湖凹陷平北北部地区平湖组储层成岩相分析[J]. 天然气地球科学, 2021, 32(7):971-981.
	ZHENG Xin, YIN Guofeng, WANG Jianwei, et al. Study on diagenetic facies of reservoir of Pinghu formation in the northern Pingbei area of the Xihu depression,East China Sea basin[J]. Natural Gas Geoscience, 2021, 32(7):971-981.
[13]	佟秀秀, 康志宏. 基于多元线性回归和BP神经网络的单井能力预测[J]. 科学技术与工程, 2019, 19(29):96-102.
	TONG Xiuxiu, KANG Zhihong. Single well capacity prediction based on multiple linear regression and back propagation neural network[J]. Science Technology and Engineering, 2019, 19(29):96-102.
[14]	连承波, 李汉林, 渠芳, 等. 基于测井资料的BP神经网络模型在孔隙度定量预测中的应用[J]. 天然气地球科学, 2006, 17(3):382-384.
	LIAN Chengbo, LI Hanlin, QU Fang, et al. Prediction of porosity based on BP artificial neural network with well logging data[J]. Natural Gas Geoscience, 2006, 17(3):382-384.
[15]	杨柳青, 查蓓, 陈伟. 基于深度神经网络的砂岩储层孔隙度预测方法[J]. 中国科技论文, 2020, 15(1):73-80.
	YANG Liuqing, ZHA Bei, CHEN Wei. Prediction method of reservoir porosity based on deep neural network[J]. China Sciencepaper, 2020, 15(1):73-80.

黏土矿物	选用模型	平均绝对误差/%	均方根误差	Pearson 相关系数
伊蒙混层	测井模型	4.26	5.131 6	0.808 9
伊蒙混层	组合模型	2.04	2.665 9	0.900 4
伊利石	测井模型	7.41	8.525 2	0.931 3
伊利石	组合模型	3.08	4.945 9	0.969 5
高岭石	测井模型	4.95	6.137 2	0.967 9
高岭石	组合模型	0.69	3.889 1	0.906 0
绿泥石	测井模型	4.75	5.582 1	0.796 9
绿泥石	组合模型	3.71	5.135 7	0.898 9

黏土矿物	测井模型平均绝对误差/%	组合模型平均绝对误差/%
伊蒙混层	3.75	2.46
伊利石	7.08	5.58
高岭石	3.58	1.88
绿泥石	4.14	3.86

黏土矿物	相对含量/%
黏土矿物	测井模型预测平均值	组合模型预测平均值	X射线衍射分析平均值
伊蒙混层	6.43	5.43	7.74
伊利石	71.58	72.70	67.77
高岭石	4.88	2.49	2.23
绿泥石	17.11	19.38	22.26