Research and Application of Attention-Enhanced DenseNet in Logging Lithology Recognition

Logging lithology identification is a key task in geological exploration, oil and gas development and groundwater resource management. Traditional methods rely on manual analysis, geological experience and rock sample analysis, which are not only time-consuming and laborious, but also susceptible to subjective judgment and data quality, resulting in uncertainty and limitations in the identification results. Currently, most lithology recognition methods use simpler machine learning algorithms and neural networks, with limited recognition accuracy, especially when dealing with complex logging data, which show poor accuracy and insufficient generalization ability, making it difficult to meet the demands of high precision and high efficiency. To address these challenges, we propose a logging lithology recognition method that combines DenseNet with Attention mechanism. The method realizes efficient feature reuse and mitigates the problem of gradient vanishing through dense connectivity, while the Attention mechanism is utilized to enhance the focus on key features and enhance the model's expressive ability and accuracy. The whole model effectively avoids the gradient vanishing problem while maintaining high accuracy and can focus on key features, thus improving the classification performance, which is especially suitable for processing complex high-dimensional data and tasks requiring fine feature recognition. Experimental results show that compared with other neural network models, the recognition accuracy of our method is 13.3% higher than that of the Transformer model and about 10.6% higher than that of the traditional convolutional neural network (CNN) model. This significant improvement not only demonstrates the superiority of combining DenseNet with the Attention mechanism, but also shows the strong potential of the method in parsing complex geologic data.