Monitoring of the Deformation of Deep Foundation Pit Using 3D Laser Scanning

Deformation monitoring of deep foundation pits is critical for ensuring construction safety. However, traditional methods (e.g., inclinometers) face inherent challenges such as limited spatial coverage (<30% in large-scale projects), low operational efficiency (requiring 2–3 times longer data acquisition than 3D scanning), and spatiotemporal discontinuity (single-point measurements fail to capture 3D dynamic deformation fields, leading to incomplete mechanical interpretations of soil–structure interactions). In contrast, 3D laser scanning provides rapid, non-contact, and high-resolution data acquisition that can capture comprehensive deformation fields over large areas. Therefore, this study proposes a novel deformation monitoring framework, aiming to expand the monitoring range and enhance the measurement accuracy. The proposed framework combines the extensive spatial coverage of 3D laser scanning with the corrective capability of a backpropagation neural network (BPNN) model. The proposed approach leverages sparse yet high-precision traditional monitoring data to train the BPNN, effectively correcting systematic deviations in laser scanning measurements caused by external disturbances and instrument errors. Validation at an active deep foundation pit site in Hangzhou reveals that the method reduces the mean absolute error (MAE) from 5.2 mm to 1.8 mm, with corrected scanning data consistency exceeding 80 percent compared to conventional monitoring measurements. This work establishes a scalable framework for deformation analysis and sets a technical benchmark for monitoring in large-scale deep foundation pit projects.