Non‐intrusive energy saving appliance recommender system for smart grid residential users

Demand side management is one of the key topics of smart grids. This study integrates the service computing paradigm in smart grid domain and proposes a demand side personalised recommendation system (PRS). The proposed PRS employs service recommendation techniques to infer residential users’ potential interests and needs on energy saving appliances, and then it recommends energy saving appliances to users, therefore potentially creating opportunities to save energy for the grid. The proposed approach starts by applying a non‐intrusive appliance load monitoring (NILM) method based on generalised particle filtering to disaggregate the end users’ household appliance utilisation profiles from the smart meter data. Based on the NILM results, several inference rules are applied to infer the preferences and energy consumption patterns, and to form the user profile . In parallel, information retrieval techniques are applied to extract keywords from the textual appliance advertisements (Ads), and to define the appliance profile . Finally, the similarity measurement method is applied to compare the user profile and appliance profile, to rank the appliance Ad, and to make the recommendations. Experiments are conducted to validate the proposed system.