Neural Model for Soil Moisture Level Determination Based on Weather Station Data Smart Campus Use Case
The human population is growing every year and naturally so is the need for resources. The most essential resource, water, is in danger of scarcity, both from pollution and increased use. The aim of this study is to reduce the usage of drinkable water in the agriculture sector with the use of Artificial Intelligence, specifically for green areas of undemanding flora (grass in front of buildings, houses, etc.). Conventional ways of irrigation for these green areas are human-operated, regulated by a scheduled timer, sensor directed, or some combination of those. Sensor-directed irrigation with the help of humans has proven to be efficient. This study will show how artificial intelligence replaces sensors and human labor. Using soil moisture sensors, and weather station data (rainfall, humidity, wind strength, wind direction, temperature), the artificial neural network is trained first to show with which data soil moisture data correlates the most, and after that with the data collected for one month is trained to know what is the relative moisture of soil based on current weather station data, so we can set the trigger for the irrigation system to start irrigating the fields. With this study, the need for human labor in means of controlling irrigation and sensor maintenance will be cut out, so a much cheaper and more efficient model for irrigation is achieved.