A Novel Compressive Sensing Approach for Optimizing Automated Monitoring of Excavation‐Induced Horizontal Displacements

Abstract

As excavation projects become increasingly complex, the demand for automated monitoring systems has risen due to their ability to provide continuous, real‐time data collection. However, traditional methods using automated inclinometers are cost‐prohibitive because they require a high number of sensors for accurate data acquisition. This study proposes a novel approach based on compressive sensing (CS) theory to interpret excavation‐induced horizontal displacement profiles using data from a reduced number of sensors. Validation with 19,311 displacement profiles from a 30.2‐m deep excavation project in Hangzhou, China, demonstrated the robustness of the method, achieving a maximum root mean square error (RMSE) of 6.42 mm (a 7.9% relative error for a maximum displacement of 80.9 mm), while reducing sensor deployment costs by a factor of 22 compared to traditional inclinometer techniques. The CS‐based approach consistently outperformed traditional regression models and proved effective across various sensor spacing scenarios. An analysis of 405,531 simulated cases provided an RMSE envelope, allowing engineers to balance accuracy and budget constraints when selecting sensor spacing. Additionally, comparative studies of sensor placement schemes revealed that while uniform spacing resulted in lower RMSE values and superior overall reconstruction, non‐uniform spacing more effectively captured maximum horizontal displacements, offering a cost‐efficient solution for applications that prioritize critical displacement monitoring.