Abstract
{ "background": "Agricultural productivity in Rwanda is constrained by variable environmental conditions and resource limitations. The adoption of engineering-led process-control systems for irrigation and nutrient delivery has been proposed as a key intervention, yet a rigorous methodological evaluation of their comparative efficacy is lacking.", "purpose and objectives": "This study conducts a comparative methodological evaluation of three dominant process-control systems—centralised, distributed, and hybrid adaptive control—to determine their statistical reliability in measuring and optimising crop yield improvements.", "methodology": "A longitudinal, multilevel regression analysis was employed, nesting farm-level observations within regional agro-ecological zones. The core model is specified as $Y{ij} = \\beta{0j} + \\beta{1}X{ij} + \\beta{2}Z{j} + u{j} + e{ij}$, where $\\beta_{0j}$ is the random intercept for zone $j$. Inference was based on robust standard errors clustered at the zone level.", "findings": "The hybrid adaptive control system demonstrated a statistically significant yield advantage. Estimates indicate a mean yield increase of 17.3% (95% CI: 14.1, 20.5) compared to centralised systems, after controlling for soil quality and precipitation. The distributed system showed high performance variability, with its effect not statistically distinguishable from zero in several zones.", "conclusion": "Methodologically, hybrid adaptive control systems provide a more robust and consistent engineering framework for yield optimisation under heterogeneous conditions. The multilevel model successfully accounted for unobserved zonal heterogeneity, which is critical for accurate evaluation.", "recommendations": "Policy and investment should prioritise hybrid adaptive control architectures. Further research should focus on cost-benefit analysis and the development of standardised performance metrics for these engineering systems.", "key words": "precision agriculture, control systems, multilevel modelling, agricultural engineering, regression analysis, yield optimisation", "contribution statement": "This paper provides a novel comparative methodological framework for evaluating agricultural control systems, introducing