Research on an intelligent method for seismic impedance inversion based on a hybrid attention mechanisms

Seismic impedance inversion is a key technique for reservoir prediction. In recent years, artificial intelligence has demonstrated great application potential in impedance inversion. However, it is often difficult to obtain large-scale, high-quality labeled data in practical applications. Therefore, intelligent inversion based on conventional supervised learning suffers from low accuracy and poor applicability when dealing with complex reservoirs characterized by rapid lithological changes, strong heterogeneity, and small thickness. This paper proposes an inversion method based on the CBAM hybrid attention mechanism. The method constructs an intelligent inversion model using a self-supervised learning framework. The model requires no labeled data for training, thereby overcoming the limitations imposed by insufficient labeled training data in practical applications. CBAM effectively enhances the model's ability to extract critical information and improves the accuracy and reliability of seismic inversion, while reducing model dependence on training set size. Application results on synthetic and field data demonstrate that the proposed method can achieve high-accuracy seismic impedance inversion and reservoir prediction in complicated geologic conditions.