A Digital Twin for Deep Vibro Ground Improvement-PPT Presentation

PPT Presentation

Pdf Summary

The paper presents a digital twin for monitoring deep vibro ground improvement, a process used to increase soil bearing capacity. Its primary aim is to detect unusual or potentially inadequate ground-improvement points, which are difficult to inspect directly underground. Missing a defective point can lead to excessive settlement, costly repairs, and liability.<br /><br />The system combines real-time, multidimensional machine sensor data with GPS locations. It transforms these data into representations of individual points, process phases, events, and segments, then calculates key performance indicators (KPIs) for penetration and compaction. Statistical outlier detection is used to identify unusual points, while visual tools help experts assess the results.<br /><br />The proposed interface includes geo-referenced KPI views to show site-wide patterns and unexecuted points; a subsurface model to distinguish broad changes in ground conditions from individual anomalies; heatmaps to compare KPIs and reveal correlations or clusters; and detailed views of individual points, including automatic phase segmentation and tools to flag and annotate concerns.<br /><br />The digital twin is intended to support evidence-based quality control, help experts explore machine data, reveal characteristics of subsurface conditions, and reduce the time required for inspections. Future development combines KPI-based analysis with unsupervised machine learning. An autoencoder can flag unusual datasets without labeled examples or ground truth. Comparing its results with KPI evaluations may identify additional anomalies for expert review and knowledge discovery.

Keywords

digital twin

deep vibro ground improvement

soil bearing capacity

ground improvement monitoring

real-time sensor data

GPS locations

key performance indicators

outlier detection

subsurface visualization

autoencoder anomaly detection