2020 Volume 74 Issue 1 Pages 265-272
This study is attempting to estimate the fresh properties （slump flow value, arrival time of slump flow 500mm, air content, etc.） aimed at the slump flow test of high-fluid concrete for the purpose of developing chemical admixtures by random forest which is a kind of machine learning. In this study, the learning starts from 60 features （explanatory variables） such as materials used, ingredients of admixtures, blending, kneading conditions, elapsed time, and their importance is evaluated. It was shown that each value of the fresh property can be estimated by 10 features. Moreover, the method of estimating the rheological constants was shown based on the past research results from the estimated slump flow and arrival time of slump flow 500mm. The estimated rheological constants can be expected to be used for the filling simulation of high-fluid concrete by numerical analysis.
