Random noise attenuation is an essential step in seismic data processing for improving seismic data quality and signal‐to‐noise ratio. We adopt an unsupervised machine learning approach to attenuate random noise via signal reconstruction strategy. This approach can be accomplished in the following steps: Firstly, we randomly mute a part of the input data of the neural network according to a certain percentage, and then the network outputs the reconstructed data influenced by this randomly mute. The objective function measures the distance between the input data and the reconstructed data. Secondly, we use the adaptive moment estimation algorithm to minimize the distance, and the network adjusts its internal parameters so that sparse representations can be captured by the multiple processing layers of the neural network. Finally, we take the same proportion of random mute on the raw seismic data which are fed to the trained neural network. Through this network, reconstruction of seismic data and attenuation of random noise are completed simultaneously. We use both synthetic and field data to testify the feasibility and applicability of the proposed method. Synthetic data experiment indicates that the proposed method achieves better denoised results than the conventional methods. Field data applications further demonstrate its superiority and practicality.

No abstract

No abstract

No abstract

In this paper, the flow field distribution of reservoir was calculated by commercial numerical simulation software, and the concept of streamline cluster was proposed. By extracting the spatial coordinates and attribute parameters of streamline particles, the vector flow field characterization parameters of streamline cluster potential, streamline cluster flow rate and oil content of streamline cluster are formed respectively. The Lorentz curve is introduced to evaluate the seepage velocity of the grid and to form a characterization method of reservoir flow heterogeneity. Based on the idea of balanced seepage field, the reconstruction methods of "building flow field", "planting flow field", "compensating flow field", "transferring flow field", "steadying flow field" and "controlling flow field" are established for different seepage field characteristic regions, and it is proved by an example. The results show that the property of streamline cluster between injection and production wells can effectively measure the horizontal and vertical development potential and displacement intensity of the reservoir, and the flow heterogeneity coefficient represents the fluid distribution difference in different development stages of the reservoir. The vector flow field characterization method is applied to a block in GD oilfield and its development status is evaluated. The vector reconstruction method established in this study is adopted to adjust the characteristics of different flow field regions. After the implementation of the adjustment scheme, the injection-production correspondence rate increased from 78.6% to 92.4%, the water drive control degree increased from 69.2% to 89.7%, and the EOR increased by 5.16%. The effect was obviously improved.