Fig 1.
Methodology for the current research.
Fig 2.
(a). Data collected from various countries from literature. (b). Number of publications with respect to a number of mix data for the past decade.
Fig 3.
Histogram for compressive strength (MPa).
Fig 4.
Histogram for cement (Kg/Cu.m), water (Kg/Cu.m) and mineral admixture (Kg/Cu.m).
Fig 5.
Histogram for aggregates (Kg/Cu.m).
Fig 6.
Histogram for chemical admixture (Kg/Cu.m).
Table 1.
K–Fold method for optimizing the data sets.
Table 2.
Chemical properties of OPC 53 grade.
Table 3.
Physical properties of manufacturing sand.
Table 4.
Physical properties of natural coarse aggregate.
Table 5.
Physical properties of RCA.
Table 6.
Physical properties of SP.
Table 7.
Mix proportion for self-compacting concrete with RCA.
Table 8.
Workability properties of SCC.
Fig 7.
(a). Sensitive assessment based on R2 for Normalization Technique 1. (b). Sensitive assessment based on R2 for Normalization Technique 2.
Fig 8.
(a). Sensitive assessment based on MSE for Normalization Technqiue 1. (b). Senstive assessment based on MSE for Normalization Technique 2.
Fig 9.
(a). Sensitive assessment based on RMSE for Normalization Technqiue 1. (b). Senstive assessment based on RMSE for Normalization Technique 2.
Table 9.
Summary of neurons in different hidden layers proposed in the literature.
Fig 10.
Impact of number of neurons on R2 value.
Fig 11.
Impact of number of neurons on MSE value.
Fig 12.
Impact of the number of neurons on RMSE value.
Fig 13.
(a). Influence of number of neurons on Layer 1 based on R2. (b). Influence of number of neurons on Layer 2 based on R2.
Fig 14.
(a). Influence of number of neurons on Layer 1 based on MSE. (b). Influence of number of neurons on Layer 2 based on MSE.
Fig 15.
(a). Influence of number of neurons on Layer 1 based RMSE. (b). Influence of the number of neurons on Layer 2 based on RMSE.
Fig 16.
Regression analysis for training and testing dataset for predicted and experimental results.
Fig 17.
Predicted compressive strength for number of neurons 9 for layer 1.
Fig 18.
(a). Predicted and actual compressive strenght values of SCC using the empolyed XG Boost learning algorithm. (b). Predicted and actual compressive strenght values of SCC using the empolyed LightGBM algorithm. (c). Predicted and actual compressive strenght values of SCC using the empolyed extra trees algorithms. (d). Predicted and actual compressive strenght values of SCC using the empolyed Random forest algorithms.
Table 10.
Evaluation metrics for machine learning models.
Table 11.
ANN weight and bias to each layer (Neuron = 9 and layer = 1).
Fig 19.
Predicted compressive strength from ANN equation and ANN models for training datasets.
Fig 20.
Predicted compressive strength from ANN equation and ANN models for testing datasets.
Fig 21.
Influence of input parameters towards the prediction of compressive strength.
Table 12.
Mix design from ANN model.
Fig 22.
Testing of fresh concrete properties from ANN mix design model and standard mix design model.
Table 13.
Fresh concrete properties of SCC.
Table 14.
Cube compressive strength at 7 and 28 days.