In the physical transport network, air and high-speed railway integration service links the two modes to improve the overall system efficiency. As a key element of transport organization, the train timetable determines the connection efficiency between trains and flights. However, conventional train timetable optimization typically targets the interests of a single stakeholder, failing to capture the coupling among enterprise operational costs, passenger travel experience, and social system benefits. To address this, this study considers three dimensions: enterprise value, passenger value, and social value. The random forest method is introduced to predict station-level enterprise value weights, characterizing the heterogeneity of different stations within the physical network, and connection-level passenger value weights, reflecting the value differences in various connection pairs in spatiotemporal coordination. On this basis, an integer programming model is developed to minimize the weighted objective function integrating the three dimensions. Numerical experiments are conducted using a real case of the high-speed railway line serving the Changbei Airport transfer hub to verify the feasibility and effectiveness of the proposed model. The results demonstrate that the proposed method can drive the system toward a more favorable equilibrium state while maintaining operational stability, effectively reducing passenger transfer time and improving connection success rates, thereby providing scientific decision support for the intermodal timetable design.