Geological fault identification is critical for gas storage safety, yet remains reliant on manual interpretation with inherent inefficiencies. While Artificial intelligent (AI) methods offer alternatives, their neglect of causal structures particularly physical mechanisms and geological confounders and severely limits sensitivity to subtle seismic signals. To overcome this, a multi-knowledge causally embedded graph representation approach for fault identification (FI) in seismic images is proposed. Utilizing geological prior knowledge, the seismic volume is first discretized into node sets through pixel-to-node transformation. To explicitly capture causal structures, a multi-relational feature computation method integrates: (a) spatial continuity constraints of faults as physical causality priors, and (b) seismic signal characteristics. Guided by these causal priors, Causally chain-graphs (CCGs) are constructed through knowledge-driven edge pruning and reinforcement, with connections constrained by pixel spatial receptive fields to reflect geological causality. These causally structured graphs subsequently undergo feature aggregation via graph convolutions, modeling non-Euclidean relationships between nodes. The pipeline evolves from low-level graphs to high-order fault representations, enabling end-to-end pixel-level identification. A gas storage plot in central China was used to verify the effectiveness of the proposed method, and the graph representation learning framework demonstrates excellent performance in pixel-level fault identification within complex geological structures. It efficiently detects the NE-SW trending fault system, including regional, block-bounding, and minor intra-block faults, achieving a mean accuracy of 93.11% and an average F1-score of 0.758. The framework significantly enhances the detection capability and noise immunity for subtle deep-seated faults (with amplitude variations <5%). It provides intelligent early-warning analysis for critical issues in gas storage operations, such as late-stage tectonic reactivation, compromised caprock integrity due to gently dipping boundary faults, and reservoir heterogeneity exacerbated by fault interactions.

Lithology identification is a critical aspect of geoenergy exploration, including geothermal energy development, gas hydrate extraction, and gas storage. In recent years, artificial intelligence techniques based on drill core images have made significant strides in lithology identification, achieving high accuracy. However, the current demand for advanced lithology identification models remains unmet due to the lack of high-quality drill core image datasets. This study successfully constructs and publicly releases the first open-source Drill Core Image Dataset (DCID), addressing the need for large-scale, high-quality datasets in lithology characterization tasks within geological engineering and establishing a standard dataset for model evaluation. DCID consists of 35 lithology categories and a total of 98,000 high-resolution images (512 × 512 pixels), making it the most comprehensive drill core image dataset in terms of lithology categories, image quantity, and resolution. This study also provides lithology identification accuracy benchmarks for popular convolutional neural networks (CNNs) such as VGG, ResNet, DenseNet, MobileNet, as well as for the Vision Transformer (ViT) and MLP-Mixer, based on DCID. Additionally, the sensitivity of model performance to various parameters and image resolution is evaluated. In response to real-world challenges, we propose a real-world data augmentation (RWDA) method, leveraging slightly defective images from DCID to enhance model robustness. The study also explores the impact of real-world lighting conditions on the performance of lithology identification models. Finally, we demonstrate how to rapidly evaluate model performance across multiple dimensions using low-resolution datasets, advancing the application and development of new lithology identification models for geoenergy exploration.