TY - GEN
T1 - Comparison between machine learning algorithms for TBM advance rate prediction
AU - Huang, Shengfeng
AU - Dastpak, Pooya
AU - Esmaeilpour, Misagh
AU - Liu, Kaijian
AU - Sousa, Rita L.
N1 - Publisher Copyright:
© 2023 The Author(s).
PY - 2023
Y1 - 2023
N2 - Prediction of tunneling performance is intimately connected to the prediction of Advance Rate (AR), which is crucial to optimize tunnelling operation and steering tunnelling processes. However, the forecasting AR is accuracy remains a challenge due to its dependency on several factors. Several methods have been introduced to predict the performance of tunnel driving, which are often of two types: closed-form equations or regression analysis. In the study, we developed regression models to predict AR from tunnel boring machine (TBM) based on machine learning (ML) algorithms. Five different popular ML algorithms, including k-nearest neighbor (KNN), support vector regression (SVR) model, artificial neural networks (ANN), random forest regression (RF), and classification and regression tree (DT), were applied to develop TBM AR prediction models using data monitored during construction. More than 600 dataset examples were collected from a TBM project between the Salgueiros and Sao Bento stations of the Porto light metro project in Portugal. 7 useful features were selected based on empiricism. 80% of the dataset were assigned for training and 20% for testing the models. The performance of developed models was evaluated and compared in terms of coefficient of determination (R2) and root mean square error (RMSE). The results show that SVR is the best AR predictor among the five algorithms tested.
AB - Prediction of tunneling performance is intimately connected to the prediction of Advance Rate (AR), which is crucial to optimize tunnelling operation and steering tunnelling processes. However, the forecasting AR is accuracy remains a challenge due to its dependency on several factors. Several methods have been introduced to predict the performance of tunnel driving, which are often of two types: closed-form equations or regression analysis. In the study, we developed regression models to predict AR from tunnel boring machine (TBM) based on machine learning (ML) algorithms. Five different popular ML algorithms, including k-nearest neighbor (KNN), support vector regression (SVR) model, artificial neural networks (ANN), random forest regression (RF), and classification and regression tree (DT), were applied to develop TBM AR prediction models using data monitored during construction. More than 600 dataset examples were collected from a TBM project between the Salgueiros and Sao Bento stations of the Porto light metro project in Portugal. 7 useful features were selected based on empiricism. 80% of the dataset were assigned for training and 20% for testing the models. The performance of developed models was evaluated and compared in terms of coefficient of determination (R2) and root mean square error (RMSE). The results show that SVR is the best AR predictor among the five algorithms tested.
UR - http://www.scopus.com/inward/record.url?scp=85160295242&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85160295242&partnerID=8YFLogxK
U2 - 10.1201/9781003348030-326
DO - 10.1201/9781003348030-326
M3 - Conference contribution
AN - SCOPUS:85160295242
SN - 9781003348030
T3 - Expanding Underground - Knowledge and Passion to Make a Positive Impact on the World- Proceedings of the ITA-AITES World Tunnel Congress, WTC 2023
SP - 2710
EP - 2716
BT - Expanding Underground - Knowledge and Passion to Make a Positive Impact on the World- Proceedings of the ITA-AITES World Tunnel Congress, WTC 2023
A2 - Anagnostou, Georgios
A2 - Benardos, Andreas
A2 - Marinos, Vassilis P.
PB - CRC Press/Balkema
T2 - ITA-AITES World Tunnel Congress, ITA-AITES WTC 2023 and the 49th General Assembly of the International Tunnelling and Underground Association, 2023
Y2 - 12 May 2023 through 18 May 2023
ER -