Title: Applying deep neural networks to forecast the inside temperature of a semi-arid greenhouse

Authors: Ahmed Badji; Abdelouahab Benseddik; Hocine Bensaha; Akila Boukhelifa; Salah Bouhoun; Badreddine Bezza; Djemoui Lalmi

Addresses: Laboratoire d'Instrumentation, Faculté de Génie Electrique, Université des Sciences et de la Technologie Houari Boumediene, BP 32 El-Alia, 16111 Bab-Ezzouar, Alger, Algérie; Unité de Recherche Appliquée en Energies Renouvelables, URAER, Centre de Développement des Energies Renouvelables, CDER, 47133, Ghardaïa, Algeria ' Unité de Recherche Appliquée en Energies Renouvelables, URAER, Centre de Développement des Energies Renouvelables, CDER, 47133, Ghardaïa, Algeria ' Unité de Recherche Appliquée en Energies Renouvelables, URAER, Centre de Développement des Energies Renouvelables, CDER, 47133, Ghardaïa, Algeria ' Laboratoire d'Instrumentation, Faculté de Génie Electrique, Université des Sciences et de la Technologie Houari Boumediene, BP 32 El-Alia, 16111 Bab-Ezzouar, Alger, Algérie ' Unité de Recherche Appliquée en Energies Renouvelables, URAER, Centre de Développement des Energies Renouvelables, CDER, 47133, Ghardaïa, Algeria ' Unité de Recherche Appliquée en Energies Renouvelables, URAER, Centre de Développement des Energies Renouvelables, CDER, 47133, Ghardaïa, Algeria ' Research Laboratory Materials, Energy Systems Technology and Environment (MESTEL), Université de Ghardaia, Rue de l'aeroport Noumerate, Ghardaia 47000, Algeria

Abstract: In this study, the air temperature inside a semi-arid greenhouse was investigated. The model database was built using greenhouse climatic data from a prototype greenhouse in Algeria's Ghardaia region. Over January month, external and internal climatic data were collected in order to develop and validate models for simulating environmental conditions inside the greenhouse, such as relative humidity (RHext), total radiation (GH), air pressure (P), and external temperature (Text). The main objective of this study is to compare three deep neural network models feed forward networks (FFN), Nonlinear auto-regressive network with exogenous inputs (NARX), and recurrent neural networks (RNN-LSTM) to see which one predicted temperature changes in the environment the best. The results showed that the NARX-predicted results agreed closely with the measurements; additionally, RNN provided satisfactory results, while FFN was the weakest of the three models.

Keywords: greenhouse; internal temperature; prediction; semi-arid climate; Ghardaia region.

DOI: 10.1504/IJSAMI.2024.139713

International Journal of Sustainable Agricultural Management and Informatics, 2024 Vol.10 No.3, pp.231 - 247

Received: 24 Feb 2023
Accepted: 17 May 2023
Published online: 05 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article