ANN Model for Energy Demand and Supply Forecasting in a Hybrid Energy Supply System

Abstract

This paper presents short term demand and supply forecasting model for a microgrid supply system used to secure the electricity demands of a commercial building, using one year demand data collected in hourly base. One-year renewable-based Micro-grid electricity supply data were produced by simulating its sub-systems (wind and PV supply systems). The Artificial Neural Network, ANN, forecasting models are built on predicting generation capacity and load demands in the next 24 hours. The ANN model presented here is a micro-level supply and demand forecasting model that links the decision making with the performance measures. To sustain the model results, the daily weather forecasts supplied by local authorities, are incorporated in our model. The models validity were tested by calculating the Mean Absolute Percent Error for the forecasted data. The ANN models' applicability and performance were tested in a case study for forecasting the demands of a hotel building and the supply potential of its microgrid supply sub-system. The building demands are assumed to be supplied by a hybrid supply system of 20% renewable-based Micro grid (10% Wind and 10% Photovoltaic) and 80% from electricity grid.