Artificial Intelligence Simulating Grain Productivity During the Wheat Development Considering Biological And Environmental Indicators

Abstract

The artificial neural networks modeling might simulate the efficiency of wheat grain yield involving biological and environmental conditions during the development cycle.  Considering the main succession systems in wheat crop in Brazil, the study aimed to adapt an artificial neural network architecture capable of predict the wheat grain productivity throughout the growth cycle, involving nitrogen and non-linearity of maximum air temperature and rainfall. The field experiment was conducted in two successions systems (soybean/wheat and maize/wheat) in 2017 and 2018, the trial design was in a randomize blocs with eight replicate in the level 0, 30, 60, and 120 kg ha^-1 N-fertilizer doses in the phenological stage of third fully expanded leaves. Every 30 day of the development cycle were obtained the biomass yield, maximum air temperature and accumulated rainfall information. The perceptron multi-layered artificial neural networks with backpropagation algorithm with network architecture 5-8-1 and 5-7-1 in soybean/wheat and maize/wheat system respectively, is able to simulate the wheat grain yield involving the nitrogen dose at top-dressing and the non-linearity of maximum air temperature and rainfall with biomass information obtained during the cycle crop.