A back-analysis method of deep excavation in soft soil based on BIM-NS-ML integrated technology

Researchers have represented an intelligent inversion framework for soil parameters in deep excavations is established by using BIM technology, finite difference method (FDM), and nondominated sorting genetic algorithm II (NSGA-II). Published in Smart Construction, this approach provide support for construction management and decision-making.

With the rapid development of the urban economy, the large-scale construction of subway transportation networks has become the key to alleviating traffic pressure. However, subway construction typically involves deep excavation, and most stations are located in busy residential and commercial areas. The surrounding buildings are densely populated, the pedestrian flow is large, and the underground pipe network is complex. Consequently, the construction is significantly restricted by the environment. This requires stricter deformation control of subway deep excavation projects, especially in soft soil areas. Soft soil has low strength, high compressibility, and strong sensitivity, which can easily lead to uneven foundation settlement and cause engineering accidents. Therefore, it is particularly important to accurately predict the ground deformation caused by deep excavation construction in soft soil and implement safety control based on the prediction results for accident prevention.

It is of great significance to use monitored data to back-analyze the soil parameters of deep excavation and guide construction. To improve the computing speed and accuracy, optimization algorithms are widely used, such as the particle swarm optimization (PSO) algorithm, evolution strategy algorithm (ESA), ant colony optimization (ACO), etc., which are combined with numerical calculations to perform parameter inverse analysis and optimization. The nondominated sorting genetic algorithm II (NSGA-II) is proposed based on the NSGA to solve multi-objective optimization problems. It has the advantages of fast running speed and good convergence, making it highly suitable for application in the fast and highly automated feedback analysis of deep excavation construction in soft soil. Traditional back analysis methods require manual modeling and repetitive complex calculations, which consume a lot of time and affect the construction progress of deep excavation projects.

Building information modeling (BIM) is not only capable of generating three-dimensional (3D) models of the design but also serves as a comprehensive information management platform tool. Its advantages in information integration and management provide a potential solution for the realization of automatic modeling technology. However, in subway construction, engineering design and numerical calculation are often two independent parts, which are difficult to integrate. To address this gap, a BIM-based back analysis method could seamlessly integrate numerical simulations (e.g., FEM/FDM) with real-time project data, enabling iterative calibration of soil properties during deep excavation.

In response to the above problems, this paper proposes a back analysis framework for soil parameters of deep excavation in soft soil based on the combination of BIM technology and ML algorithm. This framework effectively integrates the information integration function of BIM technology with the parameter inversion function of backpropagation neural network-nondominated sorting genetic algorithm II (BPNN-NSGA-II). In this framework, all calculation parameters are integrated into the BIM model of deep excavation.  Through geometric transformation and automated script execution, the BIM model automatically executes numerical simulations, conveniently populating the database with the data samples needed for surrogate model construction.  The inversion parameters are optimized and updated in the corresponding BIM model, achieving more efficient project management.

This paper "A back-analysis method of deep excavation in soft soil based on BIM-NS-ML integrated technology” was published in Smart Construction.

Jiang S, Li Z, Zhang R, Zheng J, Zhou S. A back-analysis method of deep excavation in soft soil based on BIM-NS-ML integrated technology. Smart Constr. 2025(1):0005, https://doi.org/10.55092/sc20250005.