Energy Management in Zero-Energy Building Using Neural Network Predictive Control

Abstract

In the past decade, there has been an increase in the interest of domestic electric energy from renewable energy sources without relying on the electrical grid. Zero-energy building concept has been developed to support energy, due to the high cost of electricity and the availability of the renewable energy. This paper proposes a novel technique for managing the energy in a zero-energy building using neural network predictive control (NNPC). NNPC is a combination of two techniques: 1) neural network and 2) model predictive control. The main features of NNPC are the ability to work in real time, the ability to connect to the Internet, simple controls, and disturbance reduction. In addition, the system can learn from the human behavior. To ensure better management of energy in a zero-energy building, this paper also introduces an artificial neural network-based technique for forecasting renewable energy sources, wind, and photovoltaic. This makes NNPC able to achieve maximum exploitation of forecasted renewable energy without using the grid. The system under study in this paper is a house supplied from a 1.75 kW hybrid system plus two sets of 100 Ah storage units. Wind/solar data are measured and collected from a site in Mansoura city, Delta, Egypt, for seven months. The study is conducted on the most energy-consuming day June 8, 2017.