Abstract: To overcome the limitations of static or dynamic single-index parameter-based model updating methods, this study proposes a dual-objective optimization approach for bridge model updating based on frequency and influence line compatibility. The feasibility and practicality of the proposed method were validated through experiments on a steel-concrete composite continuous girder bridge. First, modal testing under ambient excitation was conducted, and the first three natural frequencies were obtained using the Enhanced Frequency Domain Decomposition method. Subsequently, a single heavy truck was employed for idle-speed moving load testing to measure the time-history response of mid-span deflection. The vehicle-induced dynamic components and multi-axle effects were separated from the static deflection influence line using Variational Mode Decomposition and Tikhonov regularization. A joint objective function was constructed using frequencies and deflection influence lines, and key updating parameters were selected to establish a PSO-BP surrogate model. The finite element model was then updated through dual-objective optimization using the Fast Non-Dominated Sorting Genetic Algorithm with elite strategy, yielding Pareto-optimal solutions. The updated model's accuracy was verified against field-measured data. Results demonstrate that the relative error of the fundamental frequency decreased from 17.6% to 3.27%, while the overall relative error of deflection influence lines reduced from 39.7% to 16%. Notably, the peak relative error at control sections improved from 41.1% to 3.55%. The proposed dual-objective optimization method significantly enhances model accuracy, better reflecting the actual bridge state and providing a reliable foundation for refined numerical simulations.