MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202641070844 A) filed by Sr University on June 07, 2026, for Six-Layer Explainable Ai (xai) Pipeline Architecture For Transparent And Validated Crop Yield Prediction In Precision Agriculture.
Inventors include Mrs. Gunda Pranavi; and Dr. Pramoda Patro.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: ABSTRACT The present invention discloses a novel Six-Layer Explainable AI (XAI) Pipeline Architecture designed for transparent, accurate, and validated crop yield prediction in precision agriculture. The system addresses the critical limitations of conventional black-box machine learning models by integrating multi-modal data sources and providing human-interpretable explanations at every stage of the prediction process. The architecture comprises six modular layers: (1) Data Acquisition Layer that ingests heterogeneous data from satellites (NDVI, EVI indices), IoT soil and weather sensors, drones, and historical agricultural records; (2) Data Preprocessing & Integration Layer employing advanced cleaning, normalization, geospatial alignment, and fusion techniques such as Kalman filtering; (3) Feature Engineering & Selection Layer that extracts domain-specific agronomic features and applies XAI-guided feature importance ranking; (4) Model Training & Ensemble Prediction Layer utilizing hybrid ensemble models including LSTM for temporal sequences, XGBoost and Random Forest for tabular data, and CNNs for imagery analysis; (5) Explanation & Validation Layer as the core innovation, incorporating SHAP (SHapley Additive exPlanations), LIME, counterfactual analysis, and conformal prediction for uncertainty quantification and model validation; and (6) Deployment & Decision Support Layer that delivers real-time actionable recommendations through an intuitive dashboard and API interfaces suitable for edge and cloud deployment. The invention achieves superior prediction accuracy (R² 0.92) while ensuring full transparency, enabling farmers to understand the reasoning behind each yield forecast. Key novelties include the end-to-end layered XAI framework specifically tailored for agriculture, built-in validation loops against ground-truth data, and dynamic scenario modeling for climate variability. Unlike prior art systems that offer high accuracy but lack interpretability, this pipeline builds trust among end-users and supports sustainable decision-making by recommending precise interventions such as optimized irrigation, fertilizer application, and pest management. The system has been validated through extensive case studies on wheat, rice, and maize crops across diverse agro-climatic zones. It significantly reduces input waste, mitigates climate risks, and enhances food security. The invention is implemented as both a method and a computer-readable medium, making it scalable for smallholder farmers to large agribusinesses. This Six-Layer XAI Pipeline represents a transformative advancement in precision agriculture by combining predictive power with explainability and practical usability.
Disclaimer: Curated by HT Syndication.
