Multi-layer perceptron (MLP) neural network for predicting the modified compaction parameters of coarse-grained and fine-grained soils

The present study focused on developing the multi-layer perceptron neural network prediction model for the modified compaction parameters of coarse-grained and fine-grained soil. A total of 248 in situ collected soil samples were taken from the ongoing highways construction project work site for their quality control purposes. The collected soil samples were tested in the laboratory using Bureau of Indian Standard specification. Among 248 datasets, 179 datasets belong to coarse-grained soil, and the remaining 69 datasets are fit for fine-grained soil. The artificial neural network (ANN) algorithm, written in Python V3.7.9 platform, was adopted for the model development. The developed model exhibits the correlation coefficient (R) value more than 0.80 and 0.90 for coarse-grained and fine-grained soil, respectively. Additionally, the selected ANN models can predict MDD within ± 4% and ± 2% variations for coarse-grained and fine-grained soil, respectively. In contrast, OMC for both the soil can be predicted within ± 8% variations.