This paper investigates the potential of a vibration control-inspired method towards extending the fatigue life of railway steel bridges. Based on coupled thermal-mechanical and vehicle-track analysis, both the residual stresses from welding and these from traffic on the bridge are obtained. Subsequently, a multi-scale approach with a shell Finite Element (FE) model of the whole bridge and a solid FE model of its critical joints is put forward. The equation of motion is established for the controlled bridge, equipped with a Magnetorheological-Tuned Mass Damper (MR-TMD) system, while the combination of excitation, welding and control effects is practiced through own-developed packages and commercial software sub-model routines. The framework is showcased for the study of the Poyang Lake Railway Bridge in China. After obtaining the controlled stress states at the critical welded joint, the fatigue crack initial life is evaluated by using the critical plane method and the linear cumulative damage theory. Simulation results indicate that the multi-scale modelling approach followed, meets the accuracy needs for capturing the cracking process of the welded joint with high computational efficiency. The MR-TMD system, even when moderately reducing the critical joint stress amplitudes, can improve substantially the overall bridge fatigue resistance over the uncontrolled structure.