论文部分内容阅读
当前岩性油气藏勘探面对的储集体大多表现为地震微弱同相轴,在地震剖面上不容易识别,为此,将交替投影神经网络算法引入到地震资料处理中。在分析交替投影神经网络基本理论方法的基础上,研究利用该方法进行地震弱反射信号的分离,并设计实际地质模型进行算法验证,最后对东营凹陷南斜坡高8地区实际地震资料进行计算处理,结果显示,常规地震剖面上的薄砂体地震弱反射信息得到明显改善。理论模型和实际资料处理结果表明,交替投影神经网络算法是进行地震弱反射信息分离的一种有效方法。
At present, reservoirs facing the exploration of lithologic reservoirs are mostly characterized by weak seismic facies and are not easily identifiable on seismic sections. Therefore, the alternating projection neural network algorithm is introduced into seismic data processing. Based on the analysis of the basic theory of alternate projection neural network, this method is used to study the separation of seismic weak reflection signals and the actual geological model is designed to validate the algorithm. Finally, the actual seismic data in the high slope of South Slope of Dongying Depression are calculated and processed, The results show that the weak seismic reflection information of the thin sand body on the conventional seismic profile has been significantly improved. Theoretical model and actual data processing results show that the alternating projection neural network algorithm is an effective method for seismic weak reflection information separation.