Prediction Of Concrete Strength Using Artificial Neural Network

This report presents the prediction of concrete mix ratio using Artificial Neural Network mode ANNl. An artificial neural network model was developed trained and tested with 259 concrete mix data sets. These data sets were gotten from concrete companies sorted and used for which 70 15 and 15 were used for training validation and testing phases respectively. A 3-layered feed-forward neural network model with a back-propagation algorithm was adopted. Input layer comprises of 4 nodes representing the Fineness Modulus Coarse Aggregate ratio Water cement ratio and Maximum aggregate size and five output parameters which are compressive strength water content fine aggregate content coarse aggregate content and cement contents all in grams which are the expected output. The ANN model result was compared with other approach of concrete mix design and was considered adequate. The absolute error between the output from conversional mix design and the Artificial Neural Network predicted data was 0.00083. The results indicate the utility reliability and usefulness of the artificial neural network for accurately predicting concrete mix ratio.