Proactive Monitoring and Classification of Stored Grain Condition via Wireless Sensor Networks and Machine Learning Techniques

We consider the problem of proactive, 24/7 monitoring of the condition of stored grain by means of a wireless sensor network infrastructure. Temperature and moisture measurements by means of such a network infrastructure can help in accurately classifying the condition of stored grain piles. We employ machine learning techniques, and neural networks in particular, to classify the stored grain condition. Our results indicate it is possible to accurately classify the condition of stored grain with approximately 99% accuracy through this approach.