Leveraging Digital Twins and Demand Side Recommender Chatbot for Optimizing Smart Grid Energy Efficiency

Electricity consumers often face the challenge of selecting an optimal plan for saving energy. Strategic energy management and monitoring plays a key role in overcoming these challenges. Developments around Industry 5.0 powered smart grid proffers adequate solutions which allows end-consumers to monitor their energy performance towards effecting demand side recommendation services. Specific problems where end-users are likely to ignore recommended advice exists, thereby contributing to widening 'knowledge-action gap'. An ensemble of hybrid digital twins (DT) asset modelling based on ordinary differential equation (ODE) physics engine and data driven recurrent neural network (RNN) prediction approach alongside PageRank based asset behavior scoring algorithm deployed for demand side recommender and generative pre-trained transformers (GPT) based conversational chatbot technology show effectiveness in engaging and extending end-consumers interests in recommended advice. The novelty of the study lies in extending current scope of demand side recommender scheme via conversational chatbot interface for DT of electricity grid assets that better engage and monitors end-user's energy behavior while offering appropriate energy efficiency advice towards achieving energy conservation goals of smart grid consumers. Extensive experiments, including evaluation of end-user studies, revealed the effectiveness of proposed approach in terms of improved recommendation quality and user engagement towards net electricity demand reduction.