Développement d’un Modèle de Prédiction pour l’Agriculture de Précision Basé sur l’Intégration de Données Multi-Sources

Abstract

Precision agriculture is crucial for assisting farmers in addressing increasing food demands, optimizing scarce resources, and adapting to climatic challenges. This study results in intelligent application that amalgamates data from various sources — such as field parcels, meteorological information, geolocation, soil nutrients (NPK), and satellite imagery — and employs four analytical types: descriptive (to discern historical trends), diagnostic (to investigate factors affecting yields), predictive (utilizing machine learning to anticipate future yields), and prospective (to model scenarios that assist farmers in selecting appropriate strategies). The model also accounts for particular agricultural constraints, including resource availability, seasonality, financial limitations, and farmer preferences regarding crop selection and sustainable practices. The Data-Driven Approach Automation (DDAA) system utilizes machine learning and deep learning to automate tasks by replicating human actions. An essential attribute is an interactive advisor driven by a Large Language Model (LLM) that engages with farmers, providing tailored guidance in accordance with their objectives, such as optimizing yield or conserving water. The application offers precise forecasts, practical insights, and a user-friendly interface that is multilingual and accessible. This work integrates multi-source data analysis with AI-driven guidance to promote intelligent, sustainable agriculture and enhance the accessibility of advanced tools for farmers