IoT-based expert system for fault detection in Japanese Plum leaf-turgor pressure WSN

Abstract

Industry 4.0 involves the digital transformation of industrial sectors. Given the current climate change scenario and the scarcity of water in semi-arid regions, this digital transformation has to take into account the sustainable use of water. In agriculture, one of the most water-intensive sectors, to optimise the use of water, precision irrigation techniques are being applied. As a result of the digital transformation of agriculture, a key aspect for the application of these precision irrigation techniques, the crop water stress, can be predicted from a Wireless Sensor Network (WSN) of leaf-turgor pressure sensors. However, these sensors often fail, introducing errors in the data, which could lead to inaccurate application of precision irrigation techniques compromising crops and yields. So, sensor fault identification is a must. Nevertheless, sensor fault identification is a tedious and costly task that requires an expert to manually review all sensors and each of their measurements over the last 24 h. In this work, with the aim of digitally transforming this task, an IoT-based expert system is proposed. By means of a novel learning model, this system is capable of identifying sensor faults with 84.2% f1-score and 0.94 AUC ROC. Note that to train this learning model, only real-world data gathered from an experimental plot has been used. In addition, the real-world application of the IoT-based expert system in this plot is shown and discussed. Furthermore, a novel methodology that summarises the main findings and techniques applied in this study is also illustrated.