A DATA-DRIVEN APPROACH TO SOIL MOISTURE COLLECTION AND PREDICTION USING A WIRELESS SENSOR NETWORK AND MACHINE LEARNING TECHNIQUES

Abstract

Agriculture has been one among the foremost under-investigated areas in technology, and therefore the
development of exactitude Agriculture (PA) continues to be in its early stages. This paper proposes a knowledge-driven
methodology on building PA solutions for assortment and data modeling systems. Soil wetness, a key consider the crop
growth cycle, is chosen as AN example to demonstrate the effectiveness of our data-driven approach. On the gathering
aspect, a reactive wireless sensing element node is developed that aims to capture the dynamics of soil wetness
mistreatment soil wetness sensing element. The prototyped device is tested on field soil to demonstrate its practicality
and therefore the responsiveness of the sensors. On the info analysis aspect, a unique, site-specific soil wetness
prediction framework is constructed on high of models generated by the machine learning techniques SVM Regression.
The framework predicts soil wetness n days ahead supported a similar soil and environmental attributes which will be
collected by our sensing element node. Thanks to the massive knowledge size needed by the machine learning algorithms,
our framework is evaluated underneath the Illinois historical knowledge, not field collected sensing element knowledge.
It achieves low error rates (10%) and high correlations (98%) between foreseen values and actual values.