Bridge Monitoring Data Recovery Based on XGBoost-Transformer Model

To address the problem of data deficiency in bridge monitoring data, this paper combines the extreme gradient boosting algorithm XGBoost with Transformer network based on Transformer network in the context of Meixi River Bridge, constructs a combined XGBoost-Transformer model, and discusses its application characteristics, analysis process and modeling process. The analysis of the experimental results reveals that the prediction data of the XGBoost-Transformer model achieves an accuracy of 93.0% with a confidence interval of ± 0.15 mm. Compared with the Transformer model, the XGBoost-Transformer model has higher prediction accuracy and lower RMSE and MAE of 0.1014 and 0.0756, respectively, practice shows that the fusion of the time-series database InfluxDB with the XGBoost-Transformer model effectively recovers the missing bridge deflection data, increasing the utilization and analyzability of bridge deflection data.