Automated Active and Idle Time Measurement in Modular Construction Factory Using Inertial Measurement Unit and Deep Learning for Dynamic Simulation Input

Modular construction is gaining popularity in the USA for several advantages over stick-built methods in terms of reduced waste and time. However, productivity monitoring is an essential part to utilize the full potential of modular construction methods. This paper proposes a framework to automatically measure active and idle time at various workstations in modular construction factories, which essentially dictates the efficiency of production. This cycle time information can be used as inputs for dynamic prediction using simulation modeling. Vibration data were collected from workstations using inertial measurement units (IMUs), and a deep learning network was used to extract active and idle time from the vibration data. The result of this study showed that the proposed methodology can automatically calculate the active and idle time at various workstations with a 2.7% average error. This presents the potential of utilizing sensors and AI with simulation modeling for production monitoring and control.