Research on an Expressway Bridge Health Monitoring System Based on LSTM Nonlinear Time Series Modeling

Abstract With the rapid expansion of China’s expressway network, as an important infrastructure, real-time monitoring and maintenance of bridges’ health status is very important to ensure traffic safety and economic operation. Traditional bridge monitoring systems mostly adopt linear analysis methods, which makes it difficult to accurately capture the nonlinear changes of bridge structures under complex loads and environmental conditions. Therefore, this paper proposes a nonlinear time series modeling method based on long short-term memory (LSTM) network to construct a highway bridge health monitoring system. The system can accurately predict the health status and changing trend of the bridge structure through deep learning analysis of the sensor data of the bridge structure. Taking a highway bridge as an example, this paper collects vibration and strain data for 12 months and uses the LSTM model to train and test the data. The experimental results show that prediction accuracy of the model is about 15% higher than that of the traditional ARIMA model, and it has excellent performance in damage identification and structural deformation prediction, especially the ability of nonlinear feature extraction under complex working conditions. The experiment shows that the LSTM model has a root mean square error of 0.12 on a 12-month dataset, which is 42% higher than ARIMA. In the early warning task of structural damage, the success rate is 92%, and the false alarm rate is less than 5%.