Fig. 6: Proposed framework for using a kriging-based surrogate data-enriching artificial neural network (KS-ANN) for improving the prediction accuracy of the strength and coefficient of permeability of PCBM. | Nature Communications

Fig. 6: Proposed framework for using a kriging-based surrogate data-enriching artificial neural network (KS-ANN) for improving the prediction accuracy of the strength and coefficient of permeability of PCBM.

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

Fig. 6

a Preparation of PCBM samples. b Laboratory testing. c Obtain the unconfined compressive strength (\({{{{{\rm{\sigma }}}}}}\)), porosity (P), and coefficient of permeability (K) of PCBM. d Building a statistical distribution model for known data, where the focus of the color shading marker represents the statistical distribution of hydromechanical performance indicators under the combination of C and F. e Construction of Markov chain Monte Carlo (MCMC) simulations based on statistical distribution models of known data. f Enriching data within a domain using KS models. g Network structure of the ANN model. Note that F represents the compaction force, C represents the cement content, PCBM represents the permeable cement-stabilized base material, and MCMC represents the markov chain monte carlo.

Back to article page