Abstract: The transportation of large items with extremely high value, excessive weight and oversized dimensions is the highest risk category in highway freight logistics. Bridges, as critical nodes along large-cargo transportation routes, are subjected to extreme moving loads from large-cargo vehicles, making structural damage and full-state inversion urgently needed. The inverse problem of equivalent system identification for large-cargo transportation crossing a bridge is highly ill-posed. The large-cargo transportation and the bridge form a coupled dynamic system, which can be divided into a non-structural subsystem and a structural subsystem. Clearly, due to the large number of unknown state variables in the coupled system, the dimension of the parameter space to be inverted increases dramatically, ultimately leading to the unidentifiability of the bridge system. To transform the ill-posed inverse problem into a well-posed one, this study proposes a dimensionality-reduction-real-time-Bayesian method. The transformation from an ill-posed inverse problem to a well-posed one for the coupled vehicle-bridge dynamic system is achieved. By reducing the dimensionality of the non-structural subsystem, a time-varying parametric reduced-order equation for the interface forces is derived, followed by obtaining the time-varying parametric reduced-order equation for the interface forces and the bridge. Based on bridge monitoring data and real-time Bayesian method, bridge damage and full-state inversion are performed. The effectiveness of the proposed method is verified through a case study involving bridge condition monitoring during the transportation of oversized components for a nuclear power plant unit.