Prediction of fly ash concrete type using ANN and SVM models

Fly ash concrete manufactured by partly replacing cement with fly ash has evolved into higher strengths to promote large scale sustainable developments. Fly ash has been used in concretes of varying compressive strengths such as control concrete, high strength concrete (HSC), high performance concrete (HPC) and self-compacting concrete (SCC), etc. In this study, soft computing techniques such as artificial neural network (ANN) and support vector machine (SVM) are employed to predict the type of fly ash concrete among the selected concrete categories. The experimental data consisting of mix proportions of 406 nos. pertaining to control concrete, HSC, HPC and SCC are collected from literature. The models are trained with 70% of the data and remaining 30% is used for testing the trained models. The concrete ingredients such as cement, fly ash, water-binder ratio, superplasticizer, fine aggregate and coarse aggregate are used as input parameters to develop the models for prediction of the fly ash concrete type as output parameter. Statistical parameters such as correlation coefficient, mean square error, root mean square error, scatter index and objective function are used to evaluate the models’ prediction accuracy. Both the models are able to predict the fly ash concrete type with correlation above 0.97 and least errors. From the results, it is observed that ANN and SVM models have shown good capability in predicting the type of fly ash concrete which aids in designing cost effective higher concrete grades and strengths with use of local materials to satisfy specific structural and non-structural applications. For the mix proportions designed and type predicted, the application of the composite material to specific job can be defined and controlled.