Predicting freshwater production in seawater greenhouses using hybrid artificial neural network models |
|
Author: | Panahi, Fatemeh1; Ahmed, Ali Najah2; Singh, Vijay P.3; |
Organizations: |
1Faculty of Natural Resources and Earth Sciences, University of Kashan, Kashan, Iran 2Institute of Energy Infrastructure (IEI), Department of Civil Engineering, College of Engineering, Universiti Tenaga Nasional (UNITEN), 43000, Selangor, Malaysia 3Department of Biological and Agricultural Engineering, Zachry Department of Civil & Environmental Engineering, Texas A&M University, College Station, TX, USA
4Department of Water Engineering, Semnan University, Semnan, Iran
5Department of Civil Engineering, Faculty of Engineering, University of Malaya (UM), 50603, Kuala Lumpur, Malaysia 6National Water and Energy Center, United Arab Emirates University, Al Ain P.O. Box 15551, United Arab Emirates 7Water, Energy and Environmental Engineering Research Unit, University of Oulu, Finland |
Format: | article |
Version: | accepted version |
Access: | open |
Online Access: | PDF Full Text (PDF, 9.8 MB) |
Persistent link: | http://urn.fi/urn:nbn:fi-fe2021120158055 |
Language: | English |
Published: |
Elsevier,
2021
|
Publish Date: | 2021-12-01 |
Description: |
AbstractFreshwater production in seawater greenhouses (SWGH) is an important topic for decision-makers in arid lands. Since arid and semi-arid lands face water shortages, the use of SWGH helps farmers to supply water. This study proposed an integrated artificial neural network (ANN) model, namely, the ANN-antlion optimization algorithm (ANN-ALO), for predicting freshwater production in a seawater greenhouse. The width, length, and height of the evaporators and the roof transparency coefficient of the SWGH were used as the inputs of the models. The ability of ANN-ALO was benchmarked against the ANN-particle swarm optimization (ANN-PSO), ANN, and ANN-bat algorithms (ANN-BA). The novelties of the current study are the novel hybrid ANN models, the fuzzy reasoning concept for reducing the computational time, the comprehensive analysis of the uncertainty of the parameters and inputs, and the use of non-climate data. Comparing the models’ performances in the test phase demonstrated that the ANN-ALO model performed best, with a Root Mean Square Error (RMSE) value that was 18%, 33%, and 39% lower than that of the ANN-BA, ANN-PSO, and ANN models, respectively. For the ANN model, the percent bias (PBIAS) value in the training stage was 0.20, whereas for the ANN-BA, ANN-PSO, and ANN-ALO models, it was 0.14, 0.16, and 0.12, respectively. This study also indicated that the width of the seawater greenhouse was the most important parameter for predicting freshwater production. Furthermore, the results suggested that an evaporator height of 2 m resulted in the highest predicted freshwater production for all the widths except 200 m. The lowest freshwater production for different widths occurred at an evaporator height of 3 m. The generalized likelihood estimation for uncertainty analysis indicated that the uncertainty of the input parameters was lower than that of the model parameters. see all
|
Series: |
Journal of cleaner production |
ISSN: | 0959-6526 |
ISSN-E: | 1879-1786 |
ISSN-L: | 0959-6526 |
Volume: | 329 |
Article number: | 129721 |
DOI: | 10.1016/j.jclepro.2021.129721 |
OADOI: | https://oadoi.org/10.1016/j.jclepro.2021.129721 |
Type of Publication: |
A1 Journal article – refereed |
Field of Science: |
1171 Geosciences |
Subjects: | |
Copyright information: |
© 2021. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/. |
https://creativecommons.org/licenses/by-nc-nd/4.0/ |