基于叠前反演的神经网络孔隙度预测技术——以南川地区为例  

Neural Network Porosity Prediction Technology Based on Prestack Inversion——Take Nanchuan Area as an Example

作　　者：张勇[1] 何贵松 李彦婧[1] 潘兰[1] 田波[1] ZHANG Yong;HE Gui-Song;LI Yan-Jin;PAN Lan;TIAN Bo(Research Institute of Exploration and Develoment,East China Branch of SINOPEC,Nanjing 210005,China)

机构地区：[1]中国石化华东油气分公司勘探开发研究院

出　　处：《科学技术与工程》2019年第25期83-89,共7页Science Technology and Engineering

基　　金：国家科技重大专项(2016ZX05061)资助

摘　　要：孔隙度是海相页岩气富集高产一个重要因素,获取孔隙度平面特征是优选页岩储层"甜点"的一个关键环节。应用叠前反演技术及概率神经网络技术定量预测南川地区孔隙度,首先在叠前反演过程中,做好道集预处理、地震标定、子波提取,低频模型建立关键技术质量控制,获取高精度叠前反演成果;其次在概率神经网络学习训练过程中,做好交叉验证分析,优选地震属性。通过两种技术方法的结合,有效预测了南川地区孔隙度发育特征,为页岩水平井部署、钻探及区域综合评价提供资料基础。Porosity is an important factor in the high yield of marine shale gas enrichment. Obtaining the porosity plane feature is a key link in the preferred "sweet zone"of shale reservoirs. The pre-stack inversion technique and probabilistic neural network technology are used to quantitatively predict the porosity of Nanchuan area. Firstly,in the pre-stack inversion process,dodder pretreatment,seismic calibration,wavelet extraction,low frequency model establish key technical quality control,and obtain high-precision prestack inversion results. Secondly,in the process of probabilistic neural network learning and training,cross-validation analysis is performed,and seismic attributes are preferred. Through a combination of two technical methods,It effectively predicts the porosity development characteristics of Nanchuan area and provides a data basis for shale horizontal well deployment,drilling and regional comprehensive evaluation.

关 键 词：南川地区 页岩气 孔隙度 叠前反演 神经网络 

分 类 号：P631.41[天文地球—地质矿产勘探]