摘要
滩坝砂储集体具有分布较广、厚度较薄、空间分布不连续的特征。地震剖面上,通常是多个砂体以复合波的形式出现,很难形成单独的反射。滩坝砂储层信息的弱信号常被背景信息淹没,无法准确识别储层。针对滩坝砂储层的地震反射信号特点,将扩展交替投影神经网络算法引入到地震领域,对地震资料进行弱信号分离,并将算法应用到识别滩坝砂储层中,解析出砂体(组)在地震剖面上的展布特征。通过对理论模型及实际资料的试算,处理后的地震资料可以较好地展示储层展布特征,有利于滩坝砂体的识别。
Beach-bar sandstone reservoirs are characterized by wide distribution, thin thickness, and discontinuous space distribution. In seismic profile, the reservoirs including multiple sandbody often appear a form of compound wave, uneasily to form an independent reflection.We can not accurately identify them because their weak signal is often submerged into background information. So, an algorithm of extended neural network is presented to carry on the study of weak-signal separation for seismic data. And this algorithm has been applied to reservoir identification. Some distribution features of sandstones were obtained. After theoretical model and actual calculation, some processed seismic data can represent reservoir distribution features very well,which is conducive to identifying beach-bar sandbody.
出处
《天然气勘探与开发》
2015年第3期39-42,9-10,共4页
Natural Gas Exploration and Development
关键词
滩坝砂岩
神经网络
信号分离
储层预测
beach-bar sandstone,neural network,signal separation,reservoir prediction
