地震数据中杂波干扰数据的优化检测仿真研究

Optimization of Clutter Interference Data in Seismic Data Detection Simulation Research

在对地震数据中杂波干扰数据优化检测时,由于震级不同其地震数据具有一定的随机性,使得地震波产生的数据分布分散特征不易提取。传统的信号检测方法,对数据的分布规律要求较高,因此很难从分布不均匀特征数据提取合理的特征数据,导致检测不准确的问题。提出基于小波网络的地震数据中杂波干扰数据的有效排除方法,给出小波网络的结构,采用梯度下降法调整小波函数的平移因子与尺度因子,完成小波网络拟合学习。对地震数据进行加窗处理,获取加窗子样本数据矩阵,通过小波网络的训练样本数据将小波网络训练至收敛,将训练完成的小波网络作为预测模型。并循环进行莱特准则,实现杂波干扰数据的识别及排除。仿真结果表明,改进方法在排除杂波干扰数据方面具有很高的性能。

In the optimization detection of clutter interference data in seismic data,because in different seismic magnitude,the seismic data has a certain randomness,making the scattered characteristics of data distribution generated by seismic wave is not easy to extract.In traditional signal detection method,the demand of the data distribution law is higher,so it is difficult to extract the reasonable characteristics data from the uneven distribution characteristics data,leading to the problem of inaccurate detection.An effective elimination method for the clutter interference data in seismic data based on wavelet network is proposed.The structure of wavelet network is given,the gradient descent method is used to adjust the shift factor and scale factor of wavelet function to realize the fitting learning of wavelet network.Windowing processing for seismic data is made,to get the data matrix of windowing subsample,through the training sample data of wavelet network,the wavelet network is trained to convergence,the wavelet network with completed the training is as the prediction model,and to be circularly made the Letts criterion,to realize recognition and elimination of clutter interference data.The simulation results show that the improved method in the aspect of eliminating clutter interference data has very high performance.