The S-wave velocity (VS) is a fundamental parameter in geophysical analysis, prestack seismic inversion, and reservoir prediction. For various reasons, the availability of directly measured values of VS is low, especially in old wells. Therefore, indirect estimations of VS data on the basis of available reservoir information are important, and the development of a high-efficiency and low-cost prediction method is necessary. We have developed a novel prediction method that combines the convolutional neural network (CNN) and the gated recurrent unit (GRU) algorithms, based on a deep convolutional GRU (DCGRU) approach. More specifically, a CNN structure is used to identify and memorize the complex relationship between VS and well-log data, whereas a GRU network is introduced to extract key features of the data series in the depth direction. Owing to its structure, the DCGRU approach can seamlessly account for data trends with depth, local correlations across data series, and the actual depth accumulation effect. This approach is tested on data sets from an actual reservoir; it provides more reliable and accurate VS predictions not only compared with empirical models but also compared with the CNN and GRU algorithms applied separately. Our approach has potential for accurately estimating VS from log data.
9 October 2023 in IEEE Geoscience and Remote Sensing Letters
Lithology identification plays a crucial role in formation characterization and reservoir exploration. When available core samples are limited, well logs data becomes important in lithology identification. Various machine learning algorithms have been adopted to identify lithology. However, because of the spatial coupling of logging data and the vertical spatial relationship of different depths, the lithology identification of subsurface reservoirs is a challenging task. To solve this challenge, we propose a lithology identification method based on a deep learning model, which combines convolutional neural network (CNN) and long short-term memory (LSTM) network to exert their complementary advantages. In the network mentioned above, the CNN is used to extract the multiscale spatial features of the logging data, whereas the LSTM is designed to extract the vertical spatial relationship from the output features of the CNN, and finally, the mapping relationships between well logs and lithology types are established. The application results of two cases on field datasets demonstrate the effectiveness of the proposed method compared to the benchmark models. The proposed model is expected to be useful for identifying the lithology of complex strata.
19 August 2021 in Arabian Journal for Science and Engineering