Estimation of pile settlement applying hybrid radial basis function network with BBO, ALO, and GWO optimization algorithms
| Autor: | Hui Gao, Zhao Jun-Wei |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2022 |
| Předmět: | |
| Zdroj: | Journal of Applied Science and Engineering, Vol 25, Iss 6, Pp 1031-1044 (2022) |
| Druh dokumentu: | article |
| ISSN: | 2708-9967 2708-9975 |
| DOI: | 10.6180/jase.202212_25(6).0014 |
| Popis: | Deep foundations (piles) are pushed into the soil so as to carry out as permanent support of structures. Because piles can carry a large value of load, they must be precisely designed in terms of settlement. Hence, controlling and estimating of piles settlement is an important subject in pilling design and construction. The primitive objective of the present document is to discover the appropriateness of applying an optimized radial basis function neural network for foreseeing the pile settlement in rock. Here, ant lion optimization (ALO), biogeography-based optimization (BBO), and grey wolf optimization (GWO) were integrated with radial basis function (RBFNN), named ALO-RBFNN, BBO-RBFNN, and GWO-RBFNN, to determine the optimal determinative parameters of RBFNN. To use these algorithms, the results of pile driving analyzer tests and earth’s properties were measured for the Klang Valley Mass Rapid Transit (KVMRT) project built and operating in Kuala Lumpur, Malaysia. All three RBFNN models have high-level potential in the SP prediction process, in which the lowest value of R2 for the training stage is 0.9073 and 0.9015 for the testing phase. ALO-RBFNN model owns the most appropriate performance by considering coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), and variance account factor (VAF) values, which are highest in the training and testing stages. Therefore, it could be concluded that all three hybrid RBFNN models are really capable of predicting SP. However, the ALO algorithm represents a higher ability to determine the RBFNN parameters’ optimal value than other proposed algorithms. |
| Databáze: | Directory of Open Access Journals |
| Externí odkaz: |
|