Abstract

The agricultural sector has experienced increasing uncertainty due to the climate variability and soil condition fluctuations. The need for an early warning system has become essential for improving the crop resilience and sustainability. Traditional data-driven approaches have shown limitations in handling the nonlinear and uncertain nature of the soil microclimate patterns. This study has addressed the problem by proposing a Hybrid Neuro-Fuzzy Genetic Early Warning System (HNFG-EWS), which has integrated neural learning, fuzzy inference, and genetic optimization. The proposed method has combined the adaptive learning capability of neural networks with the interpretability of fuzzy systems, while genetic algorithms have optimized the rule sets and membership functions. The model has processed the soil parameters such as temperature, moisture, humidity, and pH, which have influenced the crop health. The training phase has used historical datasets, and the system has generated predictive alerts based on anomaly detection. The results demonstrate that the proposed HNFG- EWS achieves an accuracy of 91%, precision of 89%, recall of 88%, and F1-score of 89%, while maintaining an error rate of 9%. The model outperforms the neural network, fuzzy system, and genetic algorithm, which achieve lower performance values across all metrics. The system improves early detection capability, which supports reliable and timely decision-making in agricultural environments.

Authors

D.K. Mohanty¹, Thamari Thankam²
Government B.Ed. Training College Kalinga, India¹, Cihan University-Erbil, Iraq²

Keywords

Neuro-Fuzzy Systems, Genetic Algorithms, Soil Microclimate, Early Warning Systems, Precision Agriculture