Abstract: Corrosion and fatigue are key factors controlling the service safety and durability of aged and over design service life riveted steel bridges. The coupling effect of corrosion and fatigue will significantly reduce the safety of the structure and aggravate the reduction of service life. In order to evaluate the fatigue performance of aged and over design service life riveted steel bridge after corrosion, a multi-parameter pitting probability distribution and random evolution model that can reflect the corrosion law of bridge steel was established based on Markov process and probability evolution theory, and the corrosion automatic modeling program was written in Python. The corrosion damage coefficient was introduced into the classic Paris fatigue crack growth formula, and a modified Paris fatigue crack growth formula was proposed. The machine learning algorithm of support vector regression (SVR) was used to quantify the correlation between the corrosion characteristic parameters and the corrosion fatigue damage coefficient. The verification analysis results showed that the prediction error of the model was less than 5%, and the goodness of fit was 0.93. Then, the fatigue life of the riveted components after corrosion was calculated and analyzed using the modified Paris formula, and the average error between the calculated results and the physical test values was less than 10%. Through the refined simulation of the hot riveting process, the spatial transfer of the riveting force and the riveting residual stress field to the digital twin model was realized. The physical and mechanical information of the full-scale riveted joint fatigue test was mapped into the digital test to realize the data native and twin model verification. A digital twin fatigue simulation and evaluation was conducted on a corroded riveted steel plate girder bridge, and a digital fatigue twin model of the entire bridge was established, which was verified by the original data of on-site monitoring. The corrosion data was introduced into the digital fatigue twin model of the entire bridge, and the fatigue crack propagation and evolution after corrosion were simulated. The influence of corrosion parameters was analyzed to achieve data generation; the simulation results under different corrosion degrees were compared with the calculation results of the modified Paris formula, and the difference between the two was also within 10%. The digital twin fatigue evaluation model for corroded riveted steel bridges proposed in this paper quantifies the correlation between corrosion characteristic parameters and fatigue damage after corrosion, and has a certain evaluation accuracy. It can provide a reference for the service life and safety evaluation of actual engineering, and provide technical support for the construction of metaverse of bridge fatigue damage.