An Improved Random Forest–Based Operation Duration Prediction of Long-Distance Tunnel Construction Considering Geological Uncertainty
Year: 2024
Type: article
Abstract: Long-distance tunnel construction involves a sequence of construction operations that are influenced by various uncertain factors. Accurate operation duration prediction is critical to inform long-distance tunnel construction management and decision-making. In this study, a novel operation duration prediction method called random forest improved by whale optimization algorithm (WOA-RF) is proposed by considering geological conditions—the most significant uncertain factor in long-distance tunnel construction. Firstly, a geological uncertainty prediction model was established to estimate probability of geological conditions along the tunnel. Secondly, factors influencing the durations of five key construction operations, i.e., drilling, charge blasting, mucking, supporting steel frame, and shotcrete were analyzed. A prediction model for the concerned operation durations was established using the WOA-RF. Furthermore, considering the uncertainty of tunnel geological conditions, a method for calculating the expected operation duration related to a certain geological condition was proposed. Effectiveness of the proposed WOA-RF model is demonstrated in a case study, showing better performance in terms of average absolute error, root mean square error, and determination coefficient than RF model. The proposed approach can be used to inform the arrival time of the subsequent team in real time during construction, and predict the construction progress to provide a scientific basis for scheduling and timely controlling the long-distance tunnel construction progress under geological uncertainties.
Authors: Donghai Liu, Qianxin Dai, Xinlin Tang, Rui Zhang, Tingjie Lu, Junjie Chen
Institutions: Tianjin University, University of Hong Kong
Language: English
Topic: Tunneling and Rock Mechanics
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4405483511