Answers are grounded in this paper and cite their sources.
Agrivoltaic systems represent a dual-use strategy for sustainable land management, simultaneously producing renewable energy and agricultural crops. This study presents an analysis of crop yield prediction in agrivoltaic installations using ten years (2014–2024) of experimental data from Coimbatore, Tamil Nadu, India. The dataset comprises 15,840 observations across 43 variables, encompassing meteorological measurements, photovoltaic (PV) operating parameters, biomass production, and crop yields for tomato and groundnut cultivation under bifacial and semi-transparent panel configurations. Eight machine learning models were evaluated for yield prediction: Random Forest (RF), Decision Tree (DT), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Extreme Gradient Boosting (XGBoost), CatBoost, Light Gradient Boosting Machine (LightGBM), and Long Short-Term Memory (LSTM) neural networks. Results demonstrate that ensemble methods, particularly Random Forest, XGBoost, and LightGBM, achieved exceptional predictive accuracy with R² values approaching 1.0000 and mean absolute errors (MAE) below 0.06 t ha⁻¹. SHAP (SHapley Additive exPlanations) analysis revealed that aboveground biomass and seasonal harvest index were the most influential predictors of marketable yield, contributing more significantly than panel geometry or environmental variables alone. Comparative analysis across three growing seasons (Kharif, Rabi, and Summer) showed that semi-transparent panels consistently outperformed bifacial panels, with tomato yields ranging from 31–42 t ha⁻¹ versus 21–34 t ha⁻¹, respectively, during the Kharif season. This performance advantage was attributed to enhanced energy transmission. Building on this comparison, this study proposes a SHAP-Guided Stacked Ensemble–LSTM (SG-SEL) hybrid architecture that couples ensemble and recurrent learners through an interpretable stacked meta-learner. The findings provide evidence that machine learning models can predict crop performance and inform decisions regarding panel selection, irrigation strategies, and planting density optimization in dual-use agricultural systems.
No takes yet. Share an insight, caveat, or question.
Abubakar et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: