Coal seams have beddings and fissures and are typically anisotropic media. Current channel wave theories are mainly based on isotropic media, and few studies exist on the dispersion characteristics of Rayleigh channel waves in anisotropic models, such as transversely isotropic (TI) media. We chose the generalized reflection-transmission coefficient method to solve the dispersion curves of Rayleigh channel waves in TI media. However, it is difficult to solve the associated dispersion equations of Rayleigh channel waves using this method directly. Therefore, we have extended the generalized reflection-transmission coefficient method and determined the improved accuracy through numerical simulation. We analyzed the dispersion characteristics of Rayleigh channel waves of several typical coal seam models in TI media. The results showed that in the three-layer model, the difference in fundamental-mode dispersion curves between vertically transversely isotropic (VTI) media and isotropic media was relatively small; however, the differences in the higher order dispersion curves were slightly larger. The difference in the Airy phase velocity between horizontal transversely isotropic (HTI) and isotropic media was relatively large. When the coefficient of variation in the qP waves (δV) was greater than 0, the fundamental-mode and first-order phase velocity curves of HTI media exhibited an evident intersection at the head end. In the dirt-band-containing coal seam model, within the 350 and 550 Hz band, the high-frequency velocity of fundamental-mode phase velocity curve of isotropy and HTI media was slightly higher than the low-frequency velocity, which is a notable phenomenon.

The brittleness index (BI) is an important parameter for coal-bed methane (CBM) reservoir fracturing characterization. Most published studies have relied on petrophysical and well-log data to estimate the geomechanical properties of reservoir rocks. The major drawback of such methods is the lack of control away from well locations. Therefore, we have developed a method of combining BI calculation from well logs with that inverted from 3D seismic data to overcome the limitation. A real example is given here to indicate the workflow. A traditional amplitude-variation-with-offset (AVO) inversion was conducted first. BI for the CBM reservoir was then calculated from the Lamé constants inverted from prestack seismic data through a traditional AVO inversion method. We build an initial low-frequency model based on the well-log data. Comparison of the seismic inverted BI at the target reservoir and BI extracted from the well-log data showed satisfactory results. This method has been proved to be efficient and effective enough at identifying BI sweet spots in CBM reservoirs.

The coal component content is an important parameter during the coal resources exploration and exploitation. Previous logging curve regression and single neural network methods have the disadvantages of low accuracy and weak generalization ability in calculating coal component content. In this study, a gray relational analysis-hybrid neural network (GRA-HNN) method is developed by combining GRA and HNN to predict coal component content in logging data. First, the correlation degree between different conventional logging data and coal components is calculated using the GRA method, and logging curves with a correlation degree of ≥0.7 are selected as the input training data set. Then, a back propagation neural network, support vector machine neural network, and radial basis function neural network of different coal components are constructed based on the selected optimal input logging data, and the weighted average strategy is used to form an HNN prediction model. Finally, the GRA-HNN method is used to predict the coal component content of coalbed methane production wells in the Panji mining area. The application results indicate that the coal component content predicted by the GRA-HNN method has the highest accuracy compared with the logging curve regression method and its single neural network model, with a maximum average relative error of 13.4%. In addition, the accuracy of coal component content predicted by some single intelligent models is not always higher than the logging curve regression method, indicating that the neural network model is not necessarily suitable for all coal component content predictions. Our GRA-HNN method not only optimizes the prediction performance of a single neural network model by selecting effective input parameters but also comprehensively considers the prediction effect of several neural network models, which strengthens the generalization ability of neural network model and increases the log interpretation accuracy of coal component content.

The staggered grid finite difference method has emerged as one of the most commonly used approaches in finite difference methodologies due to its high computational accuracy and stability. Inevitably, discretizing over time and space domains in finite difference methods leads to numerical artifacts. This paper introduces a novel approach that combines the widely used Taylor series expansion with the least squares method to effectively suppress numerical dispersion. We have derived the coefficients for the staggered grid finite difference method by integrating Taylor series expansions with the least squares method. To validate the effectiveness of our approach, we conducted analyses on accuracy, dispersion, and stability, alongside simple and complex numerical examples. The results indicate that our method not only inherits the capabilities of the original Taylor series and least squares methods in suppressing numerical dispersion across small and medium wavenumber ranges but also surpasses the original methods. Moreover, it demonstrates robust dispersion suppression capabilities at high wavenumber ranges.