Development of the Neural Network Algorithm for the Prediction of Column Shortening in High-Rise Buildings

Abstract

The objective of this study is to propose and evaluate a neural network algorithm to predict column shortening, including drying shrinkage and creep in high-rise RC buildings. A proposed neural network algorithm for the prediction of column shortening focuses on data processing and training methods. The validity of the proposed neural network algorithm is examined through a training and prediction process based on column shortening measuring data of high-rise buildings. In the training data of a proposed neural network algorithm, the polynomial fit line of measuring data is used as the training data instead of measuring data. As a result, it has been verified that column shortening can be estimated by using the proposed neural network algorithm and that such a prediction is more accurate than what has been predicted by the conventional method using numerical models.