Prediction of concrete strength using artificial neural networks

Abstract

The purpose of this paper is to develop the I-PreConS (Intelligent PREdiction system of CONcrete Strength) that provides in-place strength information of the concrete to facilitate concrete form removal and scheduling for construction. For this purpose, the system is developed with artificial neural networks (ANN) that can learn cylinder test results as training patterns. ANN does not need a specific equation form differ from traditional prediction models. Instead of that, it needs enough input-output data. Also, it can continuously re-train the new data, so that it can conveniently adapt to new data. However the system is initially developed by the single architecture of ANN. The initial system has a problem, which it cannot appropriately predict the concrete strength when the curing temperature of a specific curing day is changed. This is because it uses the single architecture, which all nodes are fully connected, and thus it could show too plastic response. As a trial to solve this problem, modular ANN is proposed, which has multiple architectures composed of five ANNs (ANN-I~V). ANN-I predicts the early strength within 24 hours after pouring. From ANN-II to ANN-V predict the concrete strength at 2nd to 28th day after pouring. Through simulation study, the optimum architectures for individual five ANNs are determined and the best nodes are investigated for inter-connection between ANNs. Two major techniques are applied to increase the accuracy and to more precisely predict concrete strength development. One is to use parameter condensation technique in the determination of input neurons. The other is to apply the weighting technique of input neurons for more prediction accuracy. This study shows that I-PreConS using ANN is very efficient for predicting the compressive strength development of concrete.