Evolution law and influencing factors of tunnel temperature field in cold region based on data driven

To investigate the evolution law of the temperature field in cold-region tunnels and predict the temperature inside the tunnel, this study uses data processing methods to explore the distribution law of the temperature field inside cold-region tunnels, and analyzes the significant influential factors of the temperature field based on literature review. In addition, a hybrid prediction model based on XGBoost (Extreme Gradient Boosting) and LSTM (Long Short-Term Memory) is proposed to predict the temperature inside the tunnel. The results show that: The temperature inside the tunnel exhibits significant spatiotemporal coupling characteristics. The annual temperature amplitude at the tunnel portal is prominent, and the temperature distribution inside the tunnel presents two forms: symmetric (temperature difference between the entrance and exit < 2 ℃) and asymmetric (temperature difference between the entrance and exit > 2 ℃). Through data analysis and literature review, the key influential factors of the temperature field in cold-region tunnels are identified, including ambient temperature, ventilation conditions (natural wind flow, wind speed, wind direction, and the influence of operation and construction periods), and ground temperature control. Compared with traditional single-model prediction methods, the proposed hybrid prediction model based on XGBoost and LSTM has higher accuracy and strong applicability.