AI-UBREM data-driven model using neural networks for energy digital twining demonstrating EMS via solar potential prediction

Abstract In recent years, urban building energy modeling (UBEM) has been evolving using diverse technologies independent of future prediction or present retrofitting to minimize energy demand and meet generated renewable energy (RE). This research addresses UBEM combining renewable energy generation through solar potential prediction using AI data-driven digital twin model. Simultaneously, it demonstrates sustainability management through manifesting environmental management system (EMS) elements which are identifying, assessing, monitoring, and maintaining through implementing the AI-UBREM data-driven model. The model consolidates AI-based neural networks for energy demand prediction to promote building envelope retrofitting. It experimented supervised algorithms using multi-dimensional hyperparameters. DNN has successfully predicted energy demand of five building typologies with MSE loss function of 0.00694. This tool enables visualizing and predicting buildings’ energy, and solar potential towards positive energy district (PED). Consequently, diverse users can predict energy of new construction of the same five typologies. Furthermore, it provides a strategic tool to monitor and mitigate building energy consumption through retrofitting or redesign potential. AI-UBREM is an AI-data driven tool for energy and solar potential prediction that consolidates SDG 7 affordable and clean energy along with SDG 11 Sustainable cities and communities through urban energy digital twining.