In this paper focuses on crack propagation research in fatigue-critical regions such as hole edges and R-zones at stiffened panels. A simulation-based processing method for crack propagation across boundary regions is proposed, along with a Bayesian-updating-based dynamic fatigue crack propagation life prediction method. Based on the Pairs crack propagation rate model in combination with Abaqus/Zencrack, a finite element model of a scaled stiffened panel is simulated. The simulation results and the experimental data are input into a sample dataset, and a neural network is used to build a fatigue crack propagation parameter database. Dynamic Bayesian network (DBN) inference is adopted to construct a fatigue crack propagation life prediction model suitable for stiffened panels. Fatigue crack propagation experiments are made on the stiffened panel structure under condition of constant amplitude loading are also conducted. The results show that the proposed simulation processing method for crack propagation across boundar regions is enabled to effectively ensure the continuity of simulation data. The Bayesian-updating-based dynamic fatigue crack propagation life prediction method can effectively correct deviations between the simulation data and the experimental data, generating predictions still more close to the actual crack propagation behavior.