The natural phenomenon of liquid-solid two-phase flow is exceptionally common and closely related to various human production and life processes. Obtaining the flow pattern of the liquid-solid two-phase flow in pipelines is crucial for correcting real-time flow rates and ensuring the safe operation of the entire system to prevent obstructions. This study used the dynamic mode decomposition (DMD) algorithm combined with a series of preprocessing measures to reconstruct the flow pattern of a liquid-solid two-phase flow. An experimental platform was constructed to verify the results. The reconstruction effect was evaluated using the relative error. The research results indicate that although certain differences existed between the reconstructed liquid-solid two-phase flow patterns obtained by the DMD-based identification method and the true flow pattern, the differences were within a reasonable range and can objectively reflect the evolution process of the liquid-solid two-phase flow patterns. Compared with experimental and numerical simulation methods, the method for identifying liquid-solid two-phase flow patterns based on the DMD has significant advantages in terms of data processing, accuracy, efficiency, and applicability. It is not limited to specific scenarios, without the need to provide the inherent logic of data, and can simultaneously obtain modes with a single frequency and growth rate and construct reduced-order dynamic models. Thereby, it saves computational resources.