FUNDAMENTAL STUDY ON MACHINE LEARNING MODEL FOR RC BEAM SUBJECTED TO SEVERAL LOADING AND CRACK REPAIR

Abstract

Japan is prone to large earthquakes. Therefore, buildings may be subjected to multiple seismic loads. In this study, the RC cantilever specimens were subjected to shear loads three times. As the results of the test, as the number of loads increased, the crack width opened; and the stiffness decreased without repair. However, with crack repair, the crack width became small; and the stiffness recovered slightly. In addition, the machine learning model were constructed; and then it could reasonably estimate the envelope curves.