Quantification of Compaction Properties of Lateritic Soils: Usage of Hybridized ANFIS Model: Fid-Bau Portal

Quantification of Compaction Properties of Lateritic Soils: Usage of Hybridized ANFIS Model

Zhang, Jin

The material-choosing procedure for earth structures is heavily influenced by soil compaction and associated characteristics. Due to time constraints and a finished resource, there is a greater need than ever to build algorithms for forecasting compaction qualities (i.e., maximum dry unit weight (γ_dmax) as well as optimum moisture content (^ωopt) using readily observed index values. In this study, the hybrid adaptive neuro-fuzzy inference system (ANFIS) models were used to predict the γ_dmax and ^ωopt values of lateritic soils during a modified proctor compaction test. Under certain meteorological conditions, this soil might be found in subtropical and tropical regions. Two blended ANFIS frameworks were presented for the modeling technique, in which the ANFIS technique's identification factors were signaled by optimization techniques called an imperialist competitive algorithm (ICA) and whale optimization algorithm (WOA). The findings of suggested ANFIS designs for anticipating γ_dmax and ^ωopt show that both ANFISs perform admirably in the γ_dmax prediction process, with R² greater than 0.978 and 0.939, respectively, suggesting a significant correlation between simulated and real values of γ_dmax and ^ωopt. In each training and assessment set, the ANFIS optimized with the WOA (WOA-ANFIS) model performs better than the hybrid ICA-ANFIS. When the findings of this investigation are compared to the published literature, the suggested WOA-ANFIS performs much better than the published study in predicting with R² values of 0.76 and 0.707, respectively, for γ_dmax and ^ωopt. As a result, the WOA-ANFIS approach might be suggested as a framework for forecasting the compaction features of lateritic soils.