用遗传算法实现地震信号反褶积  

DECONVOLUTION OF SEISMIC SIGNAL USING GENETIC ALGORITHMS

作　　者：章珂[1] 刘贵忠[1] 

机构地区：[1]西安交通大学电信学院信息工程研究所,西安710049

出　　处：《地球物理学报》1997年第4期550-560,共11页Chinese Journal of Geophysics

摘　　要：遗传算法作为寻优手段具有全局优化和很好的稳定性．本文将遗传算法用于地震信号反褶积处理，与已往方法相比它具有更好的分辨率和稳定性我们采用Bernoulli－Gaussian模型和ARMA模型分别描述地震反射系数序列和地震子波，用最大似然和最小预测误差准则分别构造用于估计反射系数序列和地震子波的目标函数，用遗传算法优化目标函数，以实现地震信号反褶积．Global optimum and higher stability are characteristics of genetic algorithms that is a good tool for solving optimum problem. In this paper, we propose a new method of deconvolution of seismic signal based on genetic algorithms. Compared with other deconvolution methods, it has a higher resolution and higher stability. We employ Bernoulli-Gaussian model and ARMA model to represent the reflectivity sequence and seismic wavelet sequence respectively. Considering the criterion of maximum likelihood and minimum predictive error we constitute objective functions about reflectivity sequence and seismic wavelet. Each of the objective functions is optimized using genetic algorthms iteratively, which realize the blind deconvolution.

关 键 词：地震信号 反褶积 遗传算法 地震子波 地震勘探 

分 类 号：P631.443[天文地球—地质矿产勘探]