An integrated neural network for the dynamic domestic energy management of a solar house

Abstract

Energy management system in residential areas has attracted the attention of several researchers for the development of smart cities as well as smart houses. Throughout this study, a neural network based on home energy management system (NNHEMS) has been developed to set optimal consumer priorities considering many factors, such as load priority, load profiles, environmental aspects, and user comfort. This NNHEMS; has been applied to a grid-connected photovoltaic system with a storage battery to power a house. Based on the available energy of the PV system and the loads type and priority (critical and not-critical loads), the authors design an advanced control system incorporating artificial neural network (ANN) concepts in order to satisfy the energy needs of home users and increase the performance of electricity networks. In this approach, the type of the ANN is the multi-level feedback network (MLFN); interconnection is provided by the Levenberg-Marquardt algorithm; in which data and computations flow in a specific direction from input to output. The efficiency of the proposed NNHEMS; is demonstrated in an installed house in Bouismail (Algeria) during a summer week (July 2019) with favorable weather conditions. Results demonstrate that the developed NNHEMS could achieve an optimal energy management in this solar house by saving 24.6% of energy consumption. Consequently, demand and supply of renewable energy are to be improved along with electricity network efficiency increase.