• Estimation of settlement parameter of railway subgrades of high slopes underlain by low and intermediate compressible clays using an improved hybrid model ANN-IGWO. • Validation of using new experimental dataset and simulated datasets. • Perform backward parametric validation using simulated datasets. • Developed GUI framework for quick estimation of compression index. For the first time, this work develops an improved hybrid computational paradigm for the estimation and backward parametric validation of the settlement parameter of subgrade soils, i.e., compression index (C c ) of fine-grained soils, a vital parameter for assessing the settlement of pavement subgrade soils of railways, roadways, and large civil engineering constructions. The proposed method integrates an improved grey wolf optimizer (IGWO) and an artificial neural network (ANN) to develop a high-performance hybrid ANN-IGWO framework. During model construction, an improved grey wolf optimizer (IWGO) was applied to enhance the robustness of the GWO. The constructed ANN-IGWO achieved the maximum precise estimation with R 2 = 0.9524 during the validation stage. Experimental findings demonstrated that the proposed ANN-IGWO outperformed other hybrid ANNs built with six different meta-heuristic algorithms, and four standalone models, including random forests, Gaussian process regression, extreme learning machine, and deep neural network. Moreover, backward parametric validation based on the concept of parametric analysis and utilising laboratory testing results demonstrates technical robustness of the ANN-IGWO framework for future use. Overall, the developed ANN-IGWO can be employed as an efficient tool for estimating soil C c . Moreover, a graphical user interface (GUI) framework was developed for rapid scale calculation of soil C c and parametric analysis efficiently. The developed GUI and MATLAB model are provided for quick estimation of the C c of fine-grained soils utilizing basic soil parameters.
Abidhan Bardhan (Wed,) studied this question.