AGRIINTEL: A NEXT-GENERATION HYBRID ENSEMBLE INTELLIGENCE FRAMEWORK FOR PRECISION AGRICULTURE INTEGRATING INTELLIGENT CROP RECOMMENDATION AND PRECISION FERTILIZER OPTIMIZATION

Abstract

Precision agriculture is more than just a new technology— it is necessary to sustainably feed an increasing population. This will ensure food security, conserve natural resources, and promote climate-resilient agricultural systems. In this initiative, governments, agribusinesses, and farmers need to unite and work for global expansion of PA technologies for a sustainable and efficient future for agriculture. This paper presents a hybrid ensemble framework that integrates IoT-based sensor networks with machine learning for crop recommendation and fertilizer optimization. Our system combines CatBoost, XGBoost, and LightGBM algorithms to achieve high accuracy in crop suggestions and in fertilizer recommendations. The IoT simulation employs an ESP32 microcontroller interfacing with: (1) SEN0193 soil moisture sensor (0-100% range, ±3% accuracy), (2) DHT22 environmental sensor (±0.5°C temperature, ±2% humidity), and (3) JXBS-3001 NPK detector (RS485 protocol, 0-199mg/kg resolution).