Completely Automatized Recognition of the Manufacturing Process in Special Foundation ConstructionÂ-Paper

Paper

Pdf Summary

The paper presents an automated method for recognizing construction process steps from machine data, demonstrated on continuous flight auger (CFA) pile installation. Accurate, complete timing data can help identify delays and bottlenecks, but manual logging is costly and often unreliable because operators must prioritize controlling the equipment.<br /><br />The method represents each process step as a state and defines rules for transitions to the next step using machine measurements. Because steps are recognized sequentially, the system can analyze data without trying to classify the entire process at once. Process definitions and rules are kept separate from the software, making them adaptable to different equipment, site conditions, and levels of detail. The approach can also be applied retrospectively wherever suitable machine data exists.<br /><br />For the CFA example, machine readings were used to identify drilling, concreting, repositioning, tool positioning, auger cleaning, and periods of downtime. Results from eight days of production data—59 recorded hours—were compared with operator logs. Automated recognition captured short activities and interruptions that manual recording often missed. The analysis found that the machine was idle for more than one-third of the installation time, including nearly seven hours during drilling and concreting, suggesting possible delays in concrete delivery.<br /><br />The resulting timelines and summaries can support comparisons between piles, days, and sites, and enable further analysis of production rates, concrete consumption, and potential operational problems. However, machine data alone cannot reliably determine the reason for every standstill—for example, whether it reflects a break, meeting, or equipment failure—so operator input remains necessary in some cases.<br /><br />The authors conclude that rule-based recognition can provide more detailed and consistent process records than manual logging, while offering a flexible basis for identifying inefficiencies and improving construction-site management.

Keywords

construction process recognition

machine data

automated logging

CFA pile installation

state transitions

process steps

downtime analysis

rule-based recognition

production timing

construction site management