一种基于AR 模型的预测滤波压制随机噪声方法

在地震勘探中，随机噪声是一类不可避免的噪声，它的存在极大地降低了地震资料的信噪比，导致偏移成像效果差甚至不能成像。自适应线性预测滤波方法是在时间--空间域自回归( AR) 模型系数变化的假定下，把自回归( AR) 模型数学表达式进行变换，引入代价函数以提高解的稳定性、唯一性，推导得到压制一维和二维随机噪声的递归算法。通过模拟和实际地震资料验证表明，该方法能较好地压制低信噪比资料中的随机噪声干扰，同时能较好地保护有效地震信号。

A method of predictive filter pressing random noise based on AR model

Random noise is a kind of noise which is hard to advoid in seismic exploration,and it is greatly decreased signal-to-noise ratio of seismic meterials,resulting in poor or no migration imaging.Adaptive predictive filter method updated AR model formula under the assumption of AR model factors change in t-x domain,then introduced cost function to improve the stability and uniqueness of the solution and got 1D and 2D pressing noise recursive formula.Synthetic and real data examples showed that adaptive predictive filtering pressed the random noise interference in signal-to-noise ratio data and protected effective seismic wave well.