论文部分内容阅读
常规的预测反褶积算法(PD)是建立在二阶统计量的基础上,需要假设多次波和一次波是正交的,而实际数据常常不会满足这个假设。本文利用地震信号中的多次波和一次波是非高斯分布的特性,提出一种改进的基于一次波信号的非高斯性最大化的预测反褶积算法。文中将该算法简记为IPD。与常规预测反褶积算法不同的是,该算法并不需要假设多次波和一次波是正交的,仅仅假设多次波和一次波是非高斯分布的。人工合成模型和实际地震资料的处理结果表明,本文提出的基于非高斯性最大化的预测反褶积算法优于常规预测反褶积算法。
Conventional Predictive Deconvolution (PD) algorithms are based on second-order statistics. It is assumed that the multiples and the primary waves are orthogonal, and actual data often do not satisfy this assumption. In this paper, based on the non-Gaussian distribution of multiples and primary waves in seismic signals, an improved non-Gaussian maximum prediction deconvolution algorithm based on primary signals is proposed. In this paper, the algorithm is abbreviated as IPD. Unlike conventional predictive deconvolution algorithms, the algorithm does not need to assume that multiples and primaries are orthogonal, just assuming that multiples and primaries are non-Gaussian. The artificial synthetic model and the actual seismic data processing results show that the proposed non-Gaussian maximization based deconvolution algorithm is superior to the conventional prediction deconvolution algorithm.