Application of Bayesian Methods in Reliability Prediction of Truss Structures

Steel truss structures are prone to reliability degradation due to fatigue, corrosion, and other factors during long-term service. Traditional assessment methods suffer from high parameter uncertainty and costly maintenance. To improve prediction accuracy, this study proposes a Bayesian updating-based framework for truss reliability prediction, integrating ABAQUS finite element modeling with field measurements to establish a dynamic parameter updating model. Using Zhangyan Bridge as a case study, Bayesian theory is applied to fuse prior distributions with displacement monitoring data, quantifying the posterior distribution changes in member stiffness parameters under different loading modes. Results indicate that under vertical loads, the posterior standard deviation of member stiffness decreases significantly, effectively reducing parameter uncertainty. Analyses of horizontal and oblique loads further reveal the influence of load direction on reliability sensitivity. The findings provide a low-cost, practical reliability assessment method for bridge maintenance and propose future research directions, including automated algorithms and nonlinear model optimization, demonstrating substantial engineering value.