Smart irrigation through water consumption: prediction

Abstract

Irrigation is an important factor in agriculture, savingup to 50% when it is smart. This work addresses smart irrigation through an IoT prototype that uses a prediction model trained with secondary data to predict how much water to irrigate. The results showed that the best model is with TCN, achieving an R2 of 0.91 for 1 day and 0.86 for 7 days. This model is implemented in a functional prototype applied to mints that seeks to test its use in a real crop.