Internet of Things-Based Energy Consumption-Aware Framework Design for Smart Grid Environment

The widespread adoption of Internet of Things (IoT) technologies in smart grids enables fine-grained monitoring and control of energy systems. However, maintaining grid stability remains challenging when electricity production decreases unexpectedly due to fault-prone operating conditions at power generation units. This paper proposes an Artificial Intelligence of Things (AIoT)-based adaptive energy management framework that supports online adaptive demand-side control by detecting production drop anomalies and translating them into priority-aware load control actions. In practical energy systems, purely reactive strategies that trigger actions only after a demand violation may introduce temporary production–consumption imbalance and operational stress; therefore, the proposed framework targets preventive and data-driven intervention. Instead of relying on electricity production forecasting or static load shedding, the framework learns normal production behavior offline and identifies deviations using machine learning techniques. A fault modeling approach is used to generate scenario-based training data, and Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) algorithms are employed for anomaly detection and control magnitude estimation. Following anomaly identification, IoT-enabled devices are selectively regulated based on device priority levels via an MQTT-based communication infrastructure. The framework is evaluated through simulations conducted on the CupCarbon platform under normal and production degradation scenarios. Results demonstrate that early anomaly detection alone is insufficient without accurate estimation of the required demand reduction and that the proposed approach enables effective demand-side control while preserving critical loads, thereby supporting resilient smart grid operation.