Data-Driven Monitoring System for Preventing the Collapse of Scaffolding Structures

Abstract

As temporary structures, scaffolds have essential roles to hold workers, materials, and equipment throughout construction activities. However, because a safety inspection for scaffolds is primarily visual and labor intensive, the OSHA standards related to scaffolds are frequently violated. Improper management of scaffolds has caused scaffolding collapses that have a potentially detrimental effect and liability on workers’ lives. This paper discusses the significance of scaffolding collapses and explores a method to perform scaffolding monitoring. To establish an integrated method, this research cross-connects various components (e.g., strain data, finite element model (FEM)-based structural analysis, machine learning, and an actual scaffold) in the presented framework. More specifically, this framework for a smart monitoring system is involved with: (1) developing a wireless strain sensing module for data collection, (2) modeling an FEM and learning data for failure mechanisms through FEM to characterize scaffold behaviors under certain loading conditions, and (3) investigating a machine-learning algorithm (i.e., support vector machine) for decision making. The FEM simulation analyzes a scaffolding to calculate strain values for each scaffolding column from randomly generated 1,200 load cases. Load-related strain data form training and testing sets for the machine-learning algorithm that enables the distinguishing of scaffolding conditions such as safe, over-turning, uneven-settlement, and over-loading conditions. In the experimental validation, the developed wireless strain sensing modules perform the real-time strain measurement and the machine-learning algorithm to successfully estimate the status of the scaffolding structure with 97.66% accuracy on average. The proposed method could escalate a monitoring paradigm for temporary structures from a labor-intensive manual inspection to a systematic real-time approach.