Intelligent prediction method for surface settlement in tunnel excavation considering active control of shield machine parameters
To address the challenge in shield tunneling projects where reasonable shield machine parameters must be selected based on extensive engineering experience to control surface settlement, an intelligent prediction model for surface settlement in shield tunneling is proposed, considering active control of shield machine parameters. Firstly, based on the actual characteristics and spatiotem-poral features of the project, factors influencing surface settlement during shield tunneling are divided into geometric parameters, geological parameters, shield machine parameters within the tunneling section, and control parameters for the non-tunneling section. Then, tailored to the different structural characteristics of the input data, various neural network models are comprehensively applied to extract features from different input parameters, thereby constructing a hybrid neural network model that aligns with the characteristics of on-site shield tunneling data. Finally, the model's predictive perfor-mance is validated using the construction database from three sections of the Hangzhou Metro Line 3 shield tunneling project. Additionally, the proposed model is benchmarked against other traditional machine learning algorithms. The results show that compared to the random forest model and the long short-term memory neural network model, the hybrid neural network model improves the prediction performance by nearly 50％ and 30％, respectively. Among the input parameters, active control parameters of the shield machine in the non-tunneling section are identified as the most critical, exerting the highest influence on prediction outcomes. The results can provide references for settlement prediction and shield machine parameter decision-making in shield tunneling.
万方数据
维普
