An innovative GWO-BiLSTM model for predicting the advance rates of double-line subway shield tunneling with TBM

Abstract

With the growing importance of subways in public transportation, Tunnel Boring Machine (TBM) has been widely used in subway construction due to its efficiency and reliability. The Advance Rate (AR) is a key performance indicator for TBM, and accurate AR prediction is crucial for optimizing shield tunneling operations. This paper proposes a real-time AR prediction method for double-line subway projects, using data from the Shenzhen–Dayawan Intercity Line at Bainikeng Station. The method employs Wavelet Denoising (WD) for data preprocessing and develops a time series data structure scheme to enhance prediction accuracy. GWO-BiLSTM algorithm combination is first applied to tunneling prediction and benchmarked against seven conventional machine learning and deep learning algorithms. Three evaluation metrics (R², MAE, and RMSE) are used to comprehensively assess the model’s performance. The proposed method achieves an R² of 0.98022, with MAE and RMSE values of 2.1139 and 2.9527, respectively, indicating a significant improvement over other models. The improvements in data processing, time series data structuring, and algorithm integration demonstrate the superiority of the proposed method. This flexible and adaptive approach can be tailored to various geological conditions, ensuring broad applicability across different engineering contexts for effective AR prediction.