基于压缩感知的多跳地震数据采集技术与方法

随着油气地震勘探目标的复杂程度日益提高，地震数据采集系统的道容量也需要得到进一步的提升.本文根据压缩感知、稀疏表示等理论，提出了一种多跳恒传输量的数据采集框架，以减少每条测线上地震数据的传输量，进而提升采集系统的道容量.为了能够明显地提高带道能力，设计了基于有序并行原子更新的字典学习算法，该算法能够在计算量较小的前提下有效的得到相应数据的稀疏变换矩阵.基于压缩感知的多跳地震数据采集方法已能够在吉林大学研制的无缆自定位地震仪中实现.本文最后使用一组仿真数据和一组实测数据进行测试，其结果表明，数据采集信噪比控制在14 dB以上（引入噪声约18%）时，最多可以将系统的道容量提高3倍以上.

The techniques and method for multi-hop seismic data acquisition based on compressed sensing

With the increasingly complicacy in reservoirs, seismic exploration trends to be of high-density, high-fold and wide-azimuth data acquisition, which requires seismic data acquisition systems to enlarge their channel capacity. A multi-hop, constant-volume transmission seismic data acquisition method based on dictionary learning, which is according to compressed sensing and sparse representation, is developed in this work. The proposed method attempts to reduce the volume of seismic data transmitted in every line in order to increase the channel capacity of each line by times. Also, a dictionary learning algorithm based on sequenced parallel atom-update is designed, which is able to effectively and adaptively obtain the sparse land of seismic data with low computational cost. The whole method based on compressed sensing has been implemented by the cable-less self-localization seismograph which is developed by Jilin University. Synthetic and measured data are used to test this method. The results show that the compression ratio can be close to 32% when SNR is kept above 14 dB. It means that the method proposed by this paper is able to enlarge the channel capacity three times under the condition that noise is about 18 percent of signal.