Implementation of an Automated Hydroponic System using ANN: A Case Study on Spinach

The agricultural sector plays an important role in the country's economy. Hydroponics and aeroponics are growing due to modern agricultural practices which allow the cultivation of plants without soil by means of a solution containing nutrients. A certain amount of work is now being carried out in the application of artificial intelligence algorithms, such as neural networks or probabilistic networks for controlling hydroponic plant growth. The Internet of Things enables an autonomous and intelligent management of the hydroponic system by machine-to-machine interaction. This work proposes to use artificial neural networks, for the development of a smart IoT hydroponic system. The system developed is sufficiently intelligent to enable appropriate controls in the hydroponically environment, based on a number of inputs obtained. The designed system uses KNN, LSTM, RNN, SVM for the decision-making process of controlling the plants parameter growing through hydroponic techniques. The ANN, KNN, RNN, LSTM, SVM models provide the accuracy rate of 95%, 93%, 91%, 94%, 81%. However, the ANN model is preferred to train the collected data as it exhibited the best performance with 95%accuracy. Further, the macro average, a performance metric used to calculate the F1 score, yields a result of 0.94%, while the weighted average, which takes into account the number of instances associated with various class labels, results in an accuracy of 95%.