论文部分内容阅读
在岩心和录井资料较少,又非常依赖测井资料进行地质综合解释的研究区域,利用测井资料进行岩性识别是一项基础而又重要的工作.测井资料的数据种类虽然较多,但对岩性敏感的曲线较少,因此,如何优选对岩性敏感的测井曲线,然后进行网络学预测岩性,则显得尤为关键.在进行BP神经网络学习前,利用已知岩心资料,优选了本研究区对岩性较为敏感的自然伽玛和光电吸收截面指数这两种测井曲线,并做标准化与归一化处理,以消除测井系列、型号和测井曲线度量单位的不同引起的刻度和数量级误差,从而提高网络收敛速度,建立准确岩性识别模型,识别了未取芯井的岩性.研究结果表明,利用优选输入向量的BP神经网络法对苏里格气田复杂岩性进行识别,识别准确率较高,平均符合率达到了近90%.因此,通过采用该方法对岩性的识别,也为后续基础性研究工作提供了宝贵的一手资料.
In the study area where cores and log data are few and well relied on logging data for comprehensive geologic interpretation, logging data is a basic and important work to identify lithology.While logging data types are more , But less sensitive to lithology, so how to optimize the lithology-sensitive logging curve, and then predict the lithology of network is particularly crucial.Before the BP neural network learning, using known core data , The two logging curves of natural gamma ray and photoelectric absorption cross section index that are more sensitive to lithology in this study area are optimized and normalized and normalized to eliminate the need of logging series, Different induced scale and order of magnitude error, so as to improve the speed of network convergence, establish accurate lithology identification model and identify the lithology of non-coring well.The results show that the BP neural network method using the optimal input vector has complex Sulige gas field Lithology identification, identification accuracy is high, the average coincidence rate reached nearly 90%. Therefore, by using this method of identification of lithology, but also for the follow-up of basic research provides a valuable Hand information.