Development of Software to Improve Water Management in Irrigation Systems for Rice Fields Using Internet of Things and Machine Learning (#905)

Authors

Li Tang, Luis Ramón

Porta Ñaña, Diego Alexander

Delgado Vite, Jorge Luis

Abstract

Rice cultivation demands a significant amount of water for irrigation, particularly in the Piura region of Peru, where water supplies are limited. Currently, the region lacks effective systems to conserve water resources, which could increase the risk of water scarcity in the future. To address this challenge, an integrated technological solution is proposed, focusing on the development of a web and mobile application that leverages the capabilities of the Internet of Things (IoT) for device control and monitoring. This solution employs LoRa connectivity for data transmission to the receiver, which utilizes the MQTT protocol to send information to the Arduino IoT Cloud via 4G LTE. Additionally, machine learning (ML), specifically the Long Short-Term Memory (LSTM) algorithm, is integrated to predict water consumption for irrigation. These predictions are based on data from IoT devices and meteorological information obtained through the OpenWeatherMap API. The research results indicate a 56.8% reduction in water consumption when using this IoT solution compared to the traditional flood irrigation method, optimizing resource usage without impacting crop health. The water consumption prediction model, evaluated with a root mean square error (RMSE) of 1.31544, confirms the tool’s effectiveness in accurately forecasting water needs. This technological solution provides an efficient and sustainable tool to improve water management in rice cultivation in Piura and mitigate the effects of future water scarcity.