Comparison of dragonfly algorithm and Harris hawks optimization evolutionary data mining techniques for the assessment of bearing capacity of footings over two-layer foundation soils

By assist of novel evolutionary science, the classification accuracy of neural computing is improved in analyzing the bearing capacity of footings over two-layer foundation soils. To this end, Harris hawks optimization (HHO) and dragonfly algorithm (DA) are applied to a multi-layer perceptron (MLP) predictive tool for adjusting the connecting weights and biases in predicting the failure probability using seven settlement key factors, namely unit weight, friction angle, elastic modulus, dilation angle, Poisson's ratio, applied stress, and setback distance. As the first result, incorporating both HHO and DA metaheuristic algorithms resulted in higher efficiency of the MLP. Moreover, referring to the calculated area under the receiving operating characteristic curve (AUC), as well as the calculated mean square error, the DA-MLP (AUC = 0.942 and MSE = 0.1171) outperforms the HHO-MLP (AUC = 0.915 and MSE = 0.1350) and typical MLP (AUC = 0.890 and MSE = 0.1416). Furthermore, the DA surpassed the HHO in terms of time-effectiveness.

Automated summary

By utilizing novel evolutionary science, the accuracy of neural computing in analyzing the bearing capacity of footings over two-layer foundation soils is improved. The Harris hawks optimization and dragonfly algorithm are applied to a predictive tool, resulting in higher efficiency of the neural network. The dragonfly algorithm outperforms the Harris hawks optimization in terms of computational efficiency.