High-Reliability Load Recognition in Home Energy Management Systems

The increasing domestic demand for electricity has required the scientific community to develop new technologies focused on energy expenditure minimization. In this context, Home Energy Management System (HEMS) is an element that can incorporate applications aid in efficient management, such as load recognition. Load recognition allows HEMS map all appliances, assisting management. regard, methods solve problem are vital HEMS. motivation of work propose a novel system enhances applications. For purpose, proposed contains event detector stage based Wavelet transform and another appliance identification consisting two processing chains, one high reliability fast model training. main novelties our are: (i) very-low training time via chain, less than 1 second with k -Nearest Neighbor (k-NN); (ii) highest values accuracy, F 1 , kappa index, 98.90%, 98.77%, 0.9382, respectively, regularized Convolutional Neural Network (CNN-r) using ‘Personalised Retrofit Decision Support Tools UK Homes Smart Technology’ (REFIT) dataset; (iii) first paper employing CNN-r Vision Transformer (ViT) problem; (iv) approach applicable households without any restrictions, long models. Regarding tested scenarios, results reveal very strategy yielding accuracy greater 98% REFIT dataset Reference Disaggregation Dataset (REDD). Finally, presented Table 1, provides performance compared other available literature.