由于油井取芯费用高昂加之油层保护的需要 ,不可能大量取芯 ,因此利用少量的岩芯资料 ,结合其它一些手段来预测未取芯井段的岩性 ,就成为有效识别和预测潜山储层分布的重要途径之一。该文针对埕北 30太古界潜山储集体 ,运用BP神经网络技术对测井资料进行了潜山储集体岩性的定量识别 ,经与岩芯观察、薄片分析、成像测井结果对比检验 ,符合率达到了 90 % ,从而为开展潜山油藏裂缝预测研究奠定了基础。 Due to the high cost of oil well coring and the need of reservoir protection, it is impossible to coring a large amount of cores. Therefore, using a few core materials and other means to predict the lithology of non-coring wells, it becomes an effective method to identify and predict buried hill reservoirs One of the important ways of layer distribution. In this paper, according to the Archean buried-hill reservoirs in the Taobei 30, BP neural network technique was used to quantitatively identify the lithology of buried hill reservoir. Through the comparison with the core observation, the slice analysis and the imaging logging results, The coincidence rate reached 90%, which laid the foundation for the research on the prediction of fractures in buried hill reservoirs.