​

The emergence of new technologies such as Machine Learning (ML) and the Internet of Things (IoT) has revolutionized several applications and industries. Smart agriculture and irrigation are among these industries that witnessed tremendous advancement in how environmental and agricultural-related data are remotely sensed, collected, presented, and analyzed. This paper presents a smart irrigation system based on the IoT and ML to provide farmers with an efficient way to remotely sensing the environmental and agricultural-related parameters such as air temperature, pressure, humidity, and soil moisture. These data are sent wirelessly using long range wireless area networks (LoRaWAN) and stored in the cloud to be used to predict the future parameters’ values utilizing a novel enhanced wind-driven optimization-least square support vector machine (EWDO-LSSVM) algorithm. The proposed algorithm significantly improves the accuracy of predicting irrigation-related data, thus assisting farmers in making smart decisions about when and how to irrigate their farms. The presented results demonstrate that the EWDO-LSSVM outperforms both the adaptive wind-driven optimization (AWDO-LSSVM) and the original wind-driven optimization (WDO-LSSVM) algorithms, particularly in terms of the normalized root mean square error (NRMSE) and the root mean square error (RMSE) metrics. The findings of this paper reveal that EWDO-LSSVM outperformed other models with a prediction accuracy up to 87.50% for all the parameters and forecasting periods, hence making the proposed model more accurate for smart irrigation systems. © 2024 Elsevier Ltd

​