Resumo:
The agrarian sector has several applications, among them is spraying. The spraying
consists in applying phytosanitary products in the crops, for pest control and increased
production. However many times the application of plant protection products happens in
an erroneous way, having the deposition of products in unwanted places, this can occur by
drift. The drift is the dispersion of the fluids from the products caused by the air stream,
and also one of the major problems encountered in spraying. This problem also results
in socio-environmental, economic damages and an increase in the cost of production.
However there is an area of Artificial Intelligence that aims at the automatic work of
machines, machine learning, which is to make a machine learn through past experiences.
With data on drift during spraying, applied to machine learning techniques, the hypothesis
is constructed that it is possible to estimate a curve of product deposition, and present
this estimate to the user, which can correct or reduce the drift provided during spraying.
The objective of this work is to study machine learning techniques by evaluating each
technique in terms of accuracy. With the studies carried out on the techniques, it was
possible to identify that the neural network technique presented a better behavior to the
data, which obtained a better regression function
