Optimizing energy efficiency and comfort in smart homes through predictive optimization: A case study with indoor environmental parameter consideration

• Published in: Energy reports 2024-06, Vol.11, p.5619-5637
• Main Authors: Khan, Qazi Waqas, Ahmad, Rashid, Rizwan, Atif, Khan, Anam Nawaz, Lee, KyuTae, Kim, Do Hyeun
• Format: Article
• Language: English
• Subjects: Deep learning; Energy forecasting; Energy optimization; Energy price policy; Smart home
• ISSN: 2352-4847; 2352-4847
• DOI: 10.1016/j.egyr.2024.05.038

Recently, a noticeable increase in the shortage of energy resources has been observed, coupled with a rapidly escalating demand for energy. In response to this challenge, this paper proposed a dynamic predictive optimization method that considers real-time energy price policies to optimize energy usage while simultaneously ensuring occupant comfort. To this aim, real-time environment data is collected using embedded devices via sensors. Meanwhile, real-time energy usage data is collected using smart meters. The collected environment and energy data train a Deep Learning (DL) model for predicting energy and thermal comfort. The proposed framework employs a Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), and Gated Recurrent Unit (GRU) to predict humidity, temperature, Predicted Mean Vote (PMV), and energy consumption. The trained models are then deployed on the embedded devices for edge inference. The Whale Optimization Algorithm (WOA) optimizes occupant comfort and energy use. The WOA optimizes the energy cost while considering occupant comfort and real-time energy pricing. The output of whale optimization is passed to a Fuzzy Logic Controller (FLC) to trigger control commands for proactive response. The Open Connectivity Foundation (OCF) standard is used to perform the communication. The real-time OCF-based optimal actuator control testbed experiments are evaluated using several energy policies. Extensive experiments demonstrate the effectiveness of the proposed system. The developed framework achieves cost savings ranging from 35.98% to 38.22%. [Display omitted] •Build an OCF-IoTivity-based framework to optimize energy usage.•Proposed a deep learning-based model for the prediction.•A whale optimization and fuzzy logic controller are utilized for optimal control.•Six real-time energy price policies are employed to validate the performance.