Publications / 2020 Proceedings of the 37th ISARC, Kitakyushu, Japan
Recently, Off-Site Construction (OSC) is being actively applied to improve productivity by efficient factory-based production method rather than on-site production. In OSC-based construction process, problem is the accurate quality inspection for the members produced in the factory is carried out but the quality inspection for the members shipped from the factory and brought to the site is not performed properly. The existing problem in OSC-based Precast Concrete member site detection is on-site workers have to check the members directly, which is very time-consuming and expensive, and the detection accuracy is low. In addition, quality inspection is performed only in the unit of sample, not all members This study classifies the major detection items of PC members by analyzing the importance of all the detection items of PC members based on the PC member quality checklist that workers have used for on-site detection of PC members. The items that can automatically detect the damage of PC members are derived, and the types of damage necessary for the detection of each member such as deformation, crack, and wear are classified. Then, in order to apply the automatic detection technique, the data according to the damage type is collected respectively, and the damaged part is automatically detected through the machine learning. The detected damaged area is reclassified according to the degree of damage. Finally, based on the degree of damage, the status of the member is automatically identified and automatically reported to the checklist.