Hybrid Gaussian Process-Deep Learning Model for Building Energy Optimization

This article presents a research study focused on a hybrid Gaussian Process-based Deep Learning (GPDL) model designed for optimizing energy efficiency in retrofitted buildings within smart city ecosystems. The study outlines the challenges associated with traditional retrofitting methods, which often require significant time and financial resources, and can disrupt building operations. The GPDL model aims to enhance the accuracy of end-use intensity (EUI) predictions, thereby supporting better decision-making in energy management. A case study involving 6076 buildings in Mullingar City, Ireland, is detailed, where various heating, ventilation, and air conditioning (HVAC) systems were evaluated. The results indicate a substantial reduction in EUI by approximately 52.4%, leading to an overall energy savings of 39.2% for the city. The findings emphasize the importance of selecting appropriate energy systems and highlight the model's rapid prediction capabilities compared to traditional methods, making it a valuable tool for urban planners focused on developing smart cities.