Acquisition of Tower Crane Operational Performance Using Hook Mounted RTK Positioning and Image Recognition
  • Metaadatok
Tartalom: http://hdl.handle.net/10890/57723
Archívum: Műegyetem Digitális Archívum
Gyűjtemény: 1. Tudományos közlemények, publikációk
Konferenciák gyűjteményei
Creative Construction Conference
Creative Construction Conference, 2024
Cím:
Acquisition of Tower Crane Operational Performance Using Hook Mounted RTK Positioning and Image Recognition
Létrehozó:
Nishizawa, Yusuke
Mishima, Michiru
Dátum:
2024-10-07T09:06:54Z
2024-10-07T09:06:54Z
2024
Tartalmi leírás:
The data extracted from the construction sites plays a vital role in enhancing the productivity of the construction sites. The data related to the operational performance of construction equipment is one of the most important data that allows site managers to optimize the number and scheduling of equipment to reduce costs and can promptly recognize and address issues, thereby minimizing delays in the construction process. On the other hand, site managers have to manage a lot of equipment and collect the data manually, resulting in a heavy burden on them. Automated systems are currently being developed to visualize the operating performance of workbenches and forklifts. However, visualizing the operational performance of tower cranes has been an issue and challenging due to difficulties in installing sensors and extracting data from internal electronic boards. We developed a compact device that can be easily retrofitted to the crane’s hook embedded with a method to acquire operational data from the device. The device includes a camera and a GNSS receiver attached to the hook using magnets or clamps, and once powered, the captured video and positional data are transmitted directly to the cloud via LTE networks. Our method for acquiring lifting performance data consists of two steps: (1) classifying lifting operations based on the time series location data of the hook obtained through real-time kinematic (RTK) positioning and (2) identifying the materials being lifted using image recognition models like CNN based on the images taken during lifting. When our methods are applied to the data taken at different dates, our method detected lifting operations with an average accuracy of 97.7% and identified the materials being lifted with an average accuracy of 87.4%. By systematizing these methods, site managers can reduce costs, optimize crane use, and minimize delays in process workflow promptly.
Nyelv:
angol
Típus:
könyvfejezet
Formátum:
application/pdf
Azonosító: