Developing a model using neural networks to predict wheat production in the Kirkuk governorate

Abstract

Wheat is one of the most strategically essential crops in the world and one of Iraq's most important crops.The objective of the present study was to analyze energy and examine the application of a multilayer perceptron for predicting wheat yield production in the Kirkuk governorate. The research data were collected with a face-to-face inquiry made with the farmers at two fields that include the types of equipment used for the production of wheat, the number of hours worked, fuel, oil, workers, and the style of agricultural processes for the wheat crop production. The research results showed that total energy consumption in wheat was 13315.21 and 29016.27 MJha-1, while the output energy was 24867.5 and 88641 MJha-1 for the first and second fields, respectively. Seed and diesel fuel consumption are considered essential variables in wheat plantation operations, its the highest input energy values being the relative values of 30.2 and 61.97 %. These variables impacted wheat operation from 2021 to 2022 at 4020 and 17982.44 MJha-1 for the first and second fields, respectively. Finally, the results concluded that the neural network model helps predict wheat production—the neural network architecture 7-4-1 and 5-7-1 for the first and second field systems. The research shows that the trained models produced a minimum error, indicating that the test model can predict wheat yield production in the Kirkuk governorate.

Keywords

• architecture
• multilayer perceptron
• wheat
• tillage