Advances in Machine Learning for Real-Time Monitoring and Evaluation of Grain Storage Facilities: A Survey

Abstract

Grain storage loss prevention is one of the critical challenges in ensuring food security due to rising global demand and climate change. Conventional storage monitoring methods are often manual, error-prone, and inefficient and often resulting in large quantitative and qualitative losses. Rapid developments in real-time sensing technologies, Machine Learning (ML) and the Internet of Things (IoT) have enabled intelligent grain storage monitoring systems capable of detecting abnormalities, predicting spoilage, and supporting decision-making processes. This paper presents a full review of existing research on real-time grain storage monitoring using ML techniques. It categorizes and analyzes the studies based on the monitored parameters like humidity, temperature, gas emissions etc., technological frameworks such as IoT, cloud, edge computing, and applied ML algorithms like CNN, SVM, Random Forest. Strengths, limitations, and practical constraints of each approach are discussed, along with identified research gaps in data quality, system scalability, and model generalization. Furthermore, the paper explores implementation challenges in real-world scenarios and outlines future research directions such as federated learning, sensor fusion, and development of standardized datasets. The survey serves as a foundational reference for researchers, engineers, and policymakers working on smart, data-driven grain storage solutions.