Developing a digital twin framework and smart sensors gateway for monitoring soil and environmental conditions to detect anomalies and to predict crop health

We developed a portable Smart Sensor Gateway (SSG) and Digital Twin (DT) framework that integrates Sensors, Embedded Systems, Internet of Things (IoT), Artificial Intelligence (AI), and DT technologies to enhance precision in crop monitoring and management. The SSG measures critical parameters including soil moisture, temperature, Nitrogen, Phosphorus, and Potassium (NPK) levels, air conditions, and volatile organic compounds (VOCs) with data visualised through an immersive 3D digital farm representation for intuitive monitoring and anomaly tracking. Laboratory testing confirmed NPK sensor accuracy with deviations of only 2–8 kg/ha from reference values, while field experiments revealed that soil moisture significantly influences nutrient behaviour, with nitrogen showing the strongest correlation (R2 = 0.634), followed by potassium (R2 = 0.545) and phosphorus (R2 = 0.470). VOC based anomaly detection achieved 94.8% accuracy with response times under 50 ms using Isolation Forest (IF). The Autoencoder model provided complementary detection with approximately 20 ms response time, identifying 1.18% anomalies using a 95th percentile threshold. For predictive analytics, Random Forest (RF) combined with IF filtering substantially outperformed Support Vector Regression (SVR) and Multilayer Perceptron (MLP), achieving R2 values of 0.9799 for soil moisture and 0.9810 for temperature prediction. The system shows potential for 8-15% crop productivity improvements.