基于核函数主分量的维纳滤波方法研究

针对强随机噪声地震资料背景下经典维纳滤波方法在信号的保幅及高维数据空间求解过程中产生病态矩阵的问题，提出利用核函数主分量维纳滤波压制强地震勘探随机噪声.首先利用线性核函数将地震信号映射到特征空间，再通过主分量分析方法提取地震数据主分量进行数据降维，并得到核主分量维纳滤波因子，从而进行核主分量维纳滤波（K-WPC）.正演仿真及对实际地震资料处理表明，该方法对随机噪声有较好的压制作用，保幅效果也令人满意.

The research of principal-component Wiener filtering method based on kernel function

Focusing on the deficiencies in keeping amplitude of signal and generating ill-conditioned matrix when determining the solution of high-dimensional data space by the classical Wiener filtering ，this paper proposed to utilize kernel-principal-component Wiener filter(K-WPC) to process seismic data disturbed by the strong random noise. The method tansforms the seismic data to the feature space by kernel function firstly, and reduces the dimension of the transformed seismic data by principal component analysis. The filtering factors of K-WPC are then solved based on kernel matrix to execute K-WPC. The result of the simulation experiments and the application in the practical seismic data processing indicate that it can suppress the random noise and preserve the amplitude of the signal satisfactorily.