Accurately predicting airport passenger flow is crucial for optimizing operations and scheduling resources. Existing methods have limitations and are difficult to meet the needs. To this end, this paper proposes the SFusionNet model for airport passenger flow prediction. The model mines multi-dimensional segmented features of passengers, constructs accurate portraits to provide rich information; captures complex patterns and potential laws of passenger flow data with the nonlinear modeling ability of the integrated intelligent network; designs a dynamic adaptive mechanism that can automatically adjust parameters according to real-time data and changes in the operating environment. Multiple sets of experiments show that the model has better performance in various evaluation indicators and can better adapt to the dynamics and complexity of airport passenger flow.