Thesis Open Access

A NEURAL NETWORK PREDICTION MODEL TO CONTROL THE INTERIOR OF GREENHOUSES

SAMUEL AYELE AYTENFSU


Dublin Core Export

<?xml version='1.0' encoding='utf-8'?>
<oai_dc:dc xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
  <dc:creator>SAMUEL AYELE AYTENFSU</dc:creator>
  <dc:date>2019-10-25</dc:date>
  <dc:description>The main function of a greenhouse is climate adaptation by the side of the interior of the
closed structure; it aims to produce a perfect condition for the development and growth of
specific crops. It also ensures isolation and protection against harsh weather conditions
and possible animal intrusions.
This research proposed an Elman recurrent neural network prediction model to control
the interior of greenhouses by considering four input parameters that influence the
greenhouse. The input parameter considered is wind velocity,  concentration, outside
temperature and humidity. This paper also considered the data acquisition technique for
acquiring the parameter. The data used for developing the model also part of the research
paper and acquired in the place of Addis Ababa by setting experiments in property of
Ethiopian environmental and forest research institute.
The data acquisition technique is very economical and done using the Arduino platform.
The data after collected is filtered and normalized using the Min-Max normalization
technique. Then scaled all the data to {0 1} and finally, the model is developed using
Elman neural network. Several selection criteria for fixing the proper number of a hidden
neuron are analyzed; 53 different criteria is tested based on evaluating the performance
using different statistical error measurement. The design of a neural network is
implemented based on these selection criteria using a convergence theorem. After
implementation, to verify the effectiveness the proposed model simulation was conducted
in real-time inside temperature and inside relative humidity greenhouse data. The result
obtained a minimal MSE of 0.0218, MAE of 0.1016, and RMSE of 0.1477.
The proposed model is simple, economical especially for developing countries, efficient
for predicting with minimal error. These papers also include the comparison of other
models with the proposed and summarized in the table
 </dc:description>
  <dc:identifier>https://nadre.ethernet.edu.et/record/4195</dc:identifier>
  <dc:identifier>10.20372/nadre/4195</dc:identifier>
  <dc:language>eng</dc:language>
  <dc:relation>doi:10.20372/nadre/4194</dc:relation>
  <dc:relation>url:https://nadre.ethernet.edu.et/communities/aastu</dc:relation>
  <dc:relation>url:https://nadre.ethernet.edu.et/communities/nadre</dc:relation>
  <dc:rights>info:eu-repo/semantics/openAccess</dc:rights>
  <dc:rights>http://www.opendefinition.org/licenses/cc-by</dc:rights>
  <dc:subject>Greenhouse Interior Control, Elman Neural Network, Prediction Model, Convergence Theorem,</dc:subject>
  <dc:title>A NEURAL NETWORK PREDICTION MODEL TO CONTROL THE INTERIOR OF GREENHOUSES</dc:title>
  <dc:type>info:eu-repo/semantics/doctoralThesis</dc:type>
  <dc:type>publication-thesis</dc:type>
</oai_dc:dc>
13
2
views
downloads
All versions This version
Views 1313
Downloads 22
Data volume 4.1 MB4.1 MB
Unique views 77
Unique downloads 22

Share

Cite as