Fig. 4: Prediction results of the ANN/KS-ANN model. | Nature Communications

Fig. 4: Prediction results of the ANN/KS-ANN model.

From: Kriging-based surrogate data-enriching artificial neural network prediction of strength and permeability of permeable cement-stabilized base

Fig. 4

Prediction results of the KS-ANN model (a) Test dataset of different physical indicators output by the KS model. b Test dataset of mean values of different physical indicators output by the KS model. c Test dataset of the unconfined compressive strength (σ) output by the KS model. d Test dataset of porosity (P) output by the KS model. e Test dataset of the coefficient of permeability (K) output by the KS model. f Comparison between the results predicted by the KS-ANN model and the experimental results of the unconfined compressive strength (the specimens were prepared with a cement content of 10% and subjected to a compaction force of 300 kN). g Prediction results of the ANN on the test dataset from the experimental data. h Prediction results of the KS-ANN on the test dataset from the experiment. Note that ANN represents the artificial neural network, and KS-ANN represents the kriging-based surrogate data-enriching artificial neural network.

Back to article page