A drip irrigation prediction system in a greenhouse based on long short-term memory and connected objects

Abstract

Smart greenhouses use Internet of Things (IoT) technology to monitor and control various factors that affect plant growth, such as soil humidity, indoor humidity, soil temperature, rain sensor, illumination, and indoor temperature. Sensors and actuators connected to an IoT network can collect data on these factors and use it to automate processes such as watering, heating, and ventilation. This can help optimize growing conditions and improve crop yield. To enable their vegetative growth and development, plants need the right amount of water at the right time. The objective of this work is to strictly control the different factors that affect the growth of greenhouse crops. Therefore, we need a non-linear prediction model to perform greenhouse crop irrigation prediction. During operation, the system receives the input commands via sensors and then predicts the next watering run. The irrigation is predicted using GRU, LSTM, and BLSTM and a comparison was made between the results of the three techniques, and the technique with the best result was selected.