Figures & data
Table 1. The limitation of used parameters in shrinkage and creep prediction models
Table 2. The parameters and its variation range which is used for configuration of networks
Figure 1. Architecture of the primary network for assessment of the different training algorithm
Table 3. Evaluation of the different training algorithm
Figure 2. The linear regression curves of primary networks configured with different training algorithms
Table 4. Specifications of configured networks with different neurons numbers with Levenberg–Marquardt and Bayesian framework algorithms
Figure 3. The accuracy of TS10101 network using the three different data categories
Figure 4. The accuracy of RTS10101 network using the three different data categories
Table 5. The specimens mix proportions
Table 6. Distribution of the different shrinkage models error
Figure 5. The distribution of shrinkage error for different shrinkage predicting models
Figure 6. The regression curves of the neural network method and mentioned models
Table 7. Distribution of the shrinkage prediction error
Figure 7. Comparison of shrinkage strain predicted by neural networks technique with experimental observations for A300 and B300 mixtures
Figure 8. Comparison of shrinkage strain predicted by neural networks technique with experimental observations for A300-15 and B300-15 mixtures
Figure 9. Comparison of shrinkage strain predicted by neural networks technique with experimental observations for A400 and B400 mixtures
Figure 10. Comparison of shrinkage strain predicted by neural networks technique with experimental observations for A400-15 and B400-15 mixtures
