基于改进PSO-ICA的地震信号去噪方法

为了改善常规固定步长独立分量分析(ICA)算法的叠前地震信号去噪效果,本文提出一种基于混沌粒子群优化(PSO)的改进ICA算法。该算法利用混沌PSO动态调整相对梯度ICA的步长函数,减小ICA算法的稳态误差。在混沌PSO优化过程中,采用一种基于反正切函数的非线性递减惯性权重,提高PSO迭代初期的全局搜索能力和迭代后期的局部搜索能力。模型试算和实际单炮记录处理结果表明:本文提出的改进ICA算法去噪效果明显,有效信号损失小。与其他算法相比,改进ICA算法不仅能很好地保护有效地震信号,而且能提高信噪比。

Seismic denoising based on the modified particle swarm optimization-independent component analysis

A modified independent component analysis(ICA) approach based on the chaos particle swarm optimization(PSO) is proposed in this paper in order to improve prestack seismic denoising on the basis of the fixed step size ICA.This method utilizes the chaos PSO algorithm to adjust the step size function of relative gradient ICA,and it can reduce the residual error of ICA.On the other hand,a nonli-nearly decreasing inertia weight(NDIW) is proposed to improve the capability of global and local search of PSO.The trial results for synthetic and shot gather data show that the proposed ICA method works pretty well on removing random noise with very little seismic signal loses compared to other approaches.In addition,the proposed method can obtain high signal-to-noise ratio(SNR) data.