基于BP神经网络裂缝预测方法在DMT凹陷潜山变质岩中的应用研究

The Application of BP Neural Network to DMT Hill Metamorphic Fracture Prediction

裂缝油气藏在油气勘探开发中占有重要地位,准确地预测储层中的裂缝发育程度等参数是勘探开发的难点。本文采用叠后多属性分析技术对DMT潜山变质岩裂缝进行了定量预测,在储层特征分析的基础上,首先提取了对储层裂缝敏感的地震属性,包括相干体、曲率、蚂蚁体、弧长、瞬时频率、均方根振幅、反射强度属性;然后运用BP神经网络方法,对多种地震属性进行了裂缝密度的定量预测,得出裂缝发育区为东胜堡和法哈牛、静安堡构造带,次级发育区为边台、平安堡、静北、前进构造带。其结果与工区钻井资料吻合,为该区下一步的勘探开发提供了依据。

Fractured reservoirs in oil and gas exploration and development play an important role, and it is difficult to accurately predict the fractures parameters in the reservoir. In this paper, post-stack multiattribute analysis technology is used to predict quantitatively DMT hill metamorphic cracks. Based on the analysis of reservoir characteristics, sensitive seismic attributes of the reservoir fracture have been extracted, including coherent body, azimuth angle, curvature, ant body, arc length, instantaneous frequency, RMS amplitude reflection intensity, said differential, attenuation properties. Then on this basis, BP neural network is applied to predict fracture density with a variety of seismic attributes by quantitative method, and the results showed that the Dongshengbao, Fahaniu and Jing＇anbao structural zone is the fracture zone, and development area for the secondary side of this station is Biantai, Ping＇anbao and Jingbei advancing structural belt, which coincides with drilling data in the work area. Therefore, it provides the basis for the exploration and development of the next step.