April 15, 2026 | IJOCTA |
Introduction: Real-time irrigation management faces a fundamental challenge: existing frameworks typically use either static optimization models or reactive threshold-based controls, neither of which can simultaneously balance water conservation, crop yield, and environmental sustainability under dynamic conditions. Researchers from University of National and World Economy in Bulgaria addressed this gap by proposing an integrated framework that embeds a digital twin directly in a closed agent-based decision-making loop, using fuzzy multi-objective optimization to enable adaptive, real-time irrigation control validated against farm field data.
Key findings: The framework continuously updated with real soil moisture and weather sensor data, along with agent-based decision-making that follows a perception–analysis–decision–action cycle, and optimizes 4 objectives simultaneously: 1) crop growth stability, 2) water and energy minimization, 3) control fluctuation reduction, and 4) environmental compliance. Among 3 tested algorithms (NSGA-II, MOEA/D, MOPSO), NSGA-II achieved the highest hypervolume and greatest robustness across drought and extreme climate scenarios. Sensitivity analysis confirmed the framework maintains stable equilibria across parameter changes, with water consumption objectives showing the highest sensitivity. Key limitations include partial reliance on simulated data and a current scope limited to irrigation decisions, excluding fertilization and harvest timing. Future directions include multi-region field validation, extension to broader farm management decisions, and integration of adaptive learning for automatic parameter updating.
Graphical abstract
