Edge-Cloud Collaborative Event-Driven Intelligence for IoT-Based Energy Management Systems

Abstract

Off-grid photovoltaic (PV)–battery systems require adaptive control under variable load and solar conditions, where conventional threshold-based strategies remain reactive and inefficient. This paper presents an event-driven, context-aware supervisory control framework integrating IoT sensing, edge-based decision-making, and cloud-assisted contextual awareness. The approach combines real-time measurements with solar context modelling to enable anticipatory control at the edge. A Solar Recovery Indicator (SRI) is introduced to quantify near-term energy availability using current and forecast irradiance, and is integrated with a composite battery risk model to support deterministic control decisions for adaptive load prioritisation. The framework is evaluated across five operating scenarios over three experimental runs. Results show reductions of up to 2.11% in energy consumption and 1.10% in battery risk under constrained conditions, while maintaining service continuity under high load demand. Under combined stress, event trigger rates are reduced by approximately 20%, indicating improved control efficiency and reduced unnecessary actuation. The results demonstrate that embedding event-driven intelligence at the IoT edge, complemented by cloud-based contextual awareness, enables scalable, low-latency, and resilient energy management in off-grid systems.