Development of an ANN model to predict the electricity produced by small scale roof-top PV systems in Madeira Island

In this paper, an artificial neural network is proposed for estimating the electricity produced by small scale photovoltaic (PV) roof-top plants. The two main objectives of this work are: i) to determine whether the system is functioning properly and if there is a failure in the system, and ii) to estimate the magnitude of energy produced for the upcoming day by every PV system. Based on the result of this study, the energy management system can make a more efficient schedule for the use of the energy at the consumption side and monitor its own PV systems. In addition, currently, the PV system owners are notified about a failure in the system only when they receive the bill associated to the production, whereas using the proposed method conveniently would notify owners prior to bill issue. The employed artificial neural network in this work requires a number of inputs, consisting of i) the history of produced energy measurements and, ii) meteorological forecasts of solar irradiance. The comparison of the predicted value with the actual production for each day shows the validity of the method.