An AIoT-based system for real-time monitoring of tunnel construction. (March 2021)

• Record Type: Journal Article
• Authors: Zhang, Pin; Chen, Ren-Peng; Dai, Tian; Wang, Zhi-Teng; Wu, Kai
• Is Part Of: Tunnelling and underground space technology. Volume 109(2021)
• Publication Date: 2021-03
• Subjects: Internet of Things -- Artificial intelligence -- Random forest -- Shield tunnel; Tunneling -- Periodicals; Underground construction -- Periodicals; Tunnels -- Periodicals; Underground areas -- Periodicals; 624.193
• DOI: 10.1016/j.tust.2020.103766
• Languages: English
• ISSNs: 0886-7798

Abstract

Highlights: An AIoT-based system is designed to real-time control of tunnel construction information. The AIoT-based system was applied to a practical tunnel engineering. RF-based models are developed for predicting tunneling-induced settlement and shield operational parameters. Abstract: Shield machine performance and tunnelling-induced settlement are the main concerns during the tunnelling process. This study proposes an artificial intelligence Internet of Things (AIoT)-based system for real-time monitoring of tunnel construction. Shield machine operational parameters and tunnelling-induced settlement can be transferred and stored in real time by an AIoT system. Thereafter, shield operational parameters and tunnelling-induced settlement prediction models based on machine learning algorithm random forest (RF) are established based on the collected data. The models are further employed to predict shield operational parameters and ground response at the next step. This dynamic system was applied to a practical tunnel engineering. The results indicate the implementation of such process from the data collection, training and updating of RF-based models, and decision making of controlling shield machine performance can be completed within 15 minutes, which is much less than the time of excavating and installing a segmental ring, ensuring the real-time control of shield machine. Based on the predicted shield operational parameters, maximum and mean prediction error of the tunnelling-induced settlement can be controlled within 5 and 2.5 mm, respectively. The AIoT-based system improves the information and automation level during the construction process, facilitates decision-making and avoids accidents.