Smart Beehive Health Monitoring and Forecasting Based on Multi-Sensor Data and Machine Learning
Journal
2025 Asabe Annual International Meeting
Date Issued
2025
Author(s)
Abstract
Monitoring beehive health is crucial to the beekeeping industry, and the advent of Internet of Things (IoT) solutions and cloud computing has opened new avenues for enhancing colony management. By leveraging connectivity across sensor networks and cloud-based analytics, beekeepers can easily gather real-time data on colony health, detect disease outbreaks early, and promptly adjust management practices. Traditional methods of assessing beehive health often rely on manual inspections, which are time-consuming, labor-intensive, and require extensive professional expertise. Consequently, integrating automated monitoring technologies with health prediction models in a smart beehive system can substantially improve operational efficiency and mitigate the risk of hive losses. In this study, we develop an IoT-enabled health prediction system based on smart beehives, utilizing multi-sensor data collection and time series model training to forecast future trends and support data-driven decision-making. Six beehives across two different locations were instrumented with multiple key sensors to monitor temperature, humidity, weight, bee entry and exit counts, pollen collection, audio signals, and local weather conditions. These sensors connect wirelessly to a cloud platform, where the data are stored, processed, and analyzed. Our approach employs a Gated Recurrent Unit (GRU) model for forecasting three key beehive health indicators - hive weight, bee entry and exit counts, and pollen collection rate - while the traditional Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX) model is also included as a benchmark for comparison. Experimental results demonstrate the effectiveness of this IoT-based framework, as the proposed GRU model accurately predicts next-day beehive weight with a Mean Squared Error (MSE) of 0.055(kg2) and a Mean Absolute Percentage Error (MAPE) of 0.6%. These findings confirm the feasibility of applying cloud-connected sensor networks and time series modeling in the context of smart agriculture. By integrating the smart beehive system with cloud-based analytics and robust connectivity solutions, beekeepers gain access to real-time dashboards showcasing critical metrics and predictive insights into colony trends. This allows them to take timely action to adjust management methods, ultimately improving colony health, productivity, and sustainability in the broader scope of modern agricultural and natural resource management.
Subjects
beehive health prediction
machine learning
smart agriculture
time series
Publisher
American Society of Agricultural and Biological Engineers
Description
2025 American Society of Agricultural and Biological Engineers Annual International Meeting, ASABE 2025
2025 Asabe Annual International Meeting, 2025 Toronto, Ontario, Canada July 13-16, 2025
2025 Asabe Annual International Meeting, 2025 Toronto, Ontario, Canada July 13-16, 2025
Type
conference paper not in proceedings