Deep Analysis of Unbalanced Offline Data in Bridge Monitoring Based on Improved Upper Confidence Bound Algorithm

The continuous advancement of infrastructure construction and the increase in transportation pressure have caused great damage to bridge structures. Monitoring bridge structures can promptly detect potential structural damage. Therefore, this study proposes an unbalanced offline data monitoring method based on an improved upper confidence bound algorithm, aiming to provide strong support for bridge safety assessment. This study first conducts a deep analysis of the unbalanced offline data of the fusion of multi-armed bandit algorithm and upper confidence bound algorithm, then constructs an improved model, and finally analyzes the results of the proposed model. Therefore, the regret generated by the research algorithm is much smaller than that of the comparative algorithm, indicating that it can be applied to bridge monitoring and has better monitoring ability for unbalanced offline data.