古叙矿区煤体及其组合的测井曲线识别技术

不同测井曲线对于煤体结构识别具有多解性。为提高判识精度,通过对古叙矿区石宝矿段煤储层特征和常规测井响应特征分析,提取了对煤体结构反应敏感的8条测井曲线,包括自然伽马、井径Ⅰ、井径Ⅱ、深侧向电阻率、浅侧向电阻率、补偿密度、补偿中子、补偿声波,采用BP(back propagation)神经网络算法,通过MTALAB软件,建立了神经元数量为100、训练函数为TRAINLM,适应学习函数为LEARNGDM、误差分析为MSE的二层BP神经网络煤体结构定量识别模型,预测结果与矿区其他井岩心进行对比,结果表明,基于BP神经网络的煤体结构测井识别方法精确度达89%,效果好于传统的测井判识方法。

Research on Logging Curve Identification Technology of Coal Body and Combination in Guxu Mining Area

Different logging curve has multiple solutions for coal body structure identification. In order to improve the identify accuracy, according to the coal reservoirs and logging data on investigation in the Shibao block of the Guxu coal mine and extracting the 8 logging curves which are reflect sensitivity to coal body, including natural gamma ray, dual borehole diameter, deep investigate double lateral resistivity log, shallow lnvestigate double lateral resistivity, compensated density, compensated neutron and compensated sonic. Using BP neural network algorithm, the two-layer BP neural network coal structure quantitative identification model with 100 neurons, TRAINLM training function type, LEARNGDM adaptive learning function type and MSE error analysis was established by the MTALAB software. The prediction results are compared with other well cores in the mining area. The results show that the accuracy of the coal structure logging identification method based on BP neural network is 89%, which is better than the traditional logging identification method.