The tunnelling performance can only be predicted with limited reliability, since many subjective assumptions must be made (e.g., regarding the soil layers). Therefore, the production and logistic processes must be continuously measured, evaluated, and adjusted. These include the advance rate of the tunnel boring machine (TBM) and the duration of the ring construction. During the planning phase, simulation models are used to estimate the tunneling project duration depending on information from previous projects, geotechnical properties of the soil, and assumptions regarding possible delays and downtimes. The next level of deploying simulation models in the decision-making process during construction is to create real-time simulation models, which can adopt real-time data as inputs at different time points of the project to update the prediction of the performance. In this paper, we present an approach combining the benefits of the two fields of artificial intelligence and simulation modeling to create a real-time simulation model and use the real-time recorded data of a TBM to train machine-learning models. These models predict the advancing speed, ring building duration, and feed this prediction continuously to the simulation model to get the best estimation of the TBM performance in the short term and the project performance in general.
CITATION STYLE
Salloum, Y., Mahmoudi, E., & König, M. (2024). Agent-Based Simulation Model for the Real-Time Evaluation of Tunnel Boring Machines Using Deep Learning. In Lecture Notes in Civil Engineering (Vol. 390 LNCE, pp. 184–193). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-44021-2_20
Mendeley helps you to discover research relevant for your work.