Energy-aware IoT architecture for active cooling of photovoltaic panels under dynamic weather conditions

Elevated operating temperatures significantly reduce the electrical efficiency and long-term reliability of PV modules. Although active water cooling can mitigate thermal losses, its large-scale deployment is often limited by auxiliary energy consumption and the absence of energy-aware control strategies. This study presents a distributed IoT based architecture for active PV cooling, integrating autonomous ESP32-based edge nodes, a Raspberry Pi fog layer for real-time decision-making, and an optional cloud layer for long-term optimization. The system combines electrical, thermal, and meteorological measurements, including barometric pressure gradients and lightning detection, enabling adaptive control of pump activation. Experimental validation in a real photovoltaic installation demonstrated a relative daily energy gain of 7.38% for the cooled branch compared to an uncooled reference. When accounting for measured pump consumption (6 W), the resulting Energy ROI reached ROI E = 1 . 07 on representative days of high-irradiance, confirming a positive net energy balance under real operating conditions. The proposed architecture is fully wireless, scalable, and independent of centralized hardware constraints. By explicitly evaluating the net energetic effect of cooling rather than instantaneous peak gains, the study establishes a practically deployable and energetically consistent framework for adaptive PV temperature management under dynamic climatic conditions. • Distributed IoT architecture for energy-aware active cooling of photovoltaic panels. • Edge–fog separation enabling scalable sensing and centralized low-latency decision-making. • Weather-adaptive control reducing pump runtime by 18%–27% compared to temperature-only baseline. • Multi-day experimental validation (N = 52) demonstrating systematic energy gain (5.84%) and ROI E > 1 on high-irradiance days. • Scalable and extensible platform for predictive and data-driven cooling optimization.