Herding Behaviors of grasshopper and Harris hawk for hybridizing the neural network in predicting the soil compression coefficient

摘要

This work deals with proposing two novel predictors of soil compression coefficient (SCC) through hybridizing the artificial neural network (ANN) by using grasshopper optimization algorithm (GOA) and Harris hawks optimization (HHO) metaheuristic techniques. The SCC is considered as a function of twelve key factors of the soil, collocated from a local project at Hai Phong city (Vietnam). After creating the HHO-ANN and GOA-ANN ensembles, the best structure of them is determined by a sensitivity analysis process. Each model predicted the SCC and comparing the responses with target values revealed that both metaheuristic algorithms enhance the accuracy of the ANN. In details, applying the GOA and HHO resulted in reducing the ANN leaning error (root mean square error) by 14.96% and 10.88%, as well as the prediction error by 7.14% and 4.76%, respectively. (C) 2019 Elsevier Ltd. All rights reserved.