Speaker
Description
The rapid adoption of smart classrooms has significantly increased the number of connected electronic devices, resulting in higher energy consumption and greater operational complexity. Efficient energy management requires not only accurate prediction models but also reliable sensing technologies capable of continuously capturing real-time environmental and electrical conditions. Recent advances in nano-sensor technology have enabled highly sensitive, low-power, and miniaturized sensing devices that provide fine-grained monitoring for intelligent Internet of Things (IoT) environments. However, most existing energy management systems either focus solely on prediction accuracy or lack interpretable decision-support mechanisms for practical deployment.
This paper proposes a hybrid soft computing framework that integrates nano-sensor-enabled IoT monitoring, Long Short-Term Memory (LSTM), and a Mamdani Fuzzy Inference System (FIS) for predictive energy monitoring in smart classrooms. In the proposed architecture, distributed nano-sensor networks continuously acquire energy-related information, including electrical power consumption and environmental conditions such as temperature, humidity, illumination, and occupancy. The collected data are transmitted through an IoT gateway and processed by an LSTM model to capture temporal consumption patterns and forecast future energy demand. The predicted energy values are subsequently interpreted using fuzzy logic to generate human-readable energy efficiency assessments and operational recommendations.
The proposed framework was evaluated using the publicly available CU-BEMS smart building dataset, which serves as a representative benchmark for nano-sensor-based energy monitoring environments. Experimental results demonstrate excellent predictive performance, achieving a Mean Absolute Error (MAE) of 27.52 W, Root Mean Square Error (RMSE) of 50.03 W, Mean Absolute Percentage Error (MAPE) of 0.99%, and a coefficient of determination (R²) of 0.99. These results indicate that the proposed hybrid framework accurately captures temporal energy consumption patterns while providing interpretable decision support for energy optimization. The integration of nano-sensor networks with hybrid LSTM-Fuzzy intelligence offers a scalable and explainable solution for predictive energy management in smart classrooms, supporting sustainable educational facilities and next-generation intelligent building ecosystems.
