Search documents

Browse topics

Document details

Modelling and Forecasting PV Production in the Absence of Behind-the-Meter Measurements
T. Landelius, S. Andersson, R. Abrahamsson
Solar Radiation, Photovoltaic (PV), Modelling / Modeling
PV Systems - Performance, Applications and Integration
Subtopic: Solar Resource and Forecasting
Event: 35th European Photovoltaic Solar Energy Conference and Exhibition
Session: 6DO.11.2
1684 - 1689
ISBN: 3-936338-50-7
Paper DOI: 10.4229/35thEUPVSEC20182018-6DO.11.2
0,00 EUR
Document(s): paper


This paper deals with the prediction of the net load from photovoltaic installations. The forecasts were only based on information from a numerical weather prediction model and measured net load. The study thus tackles the problem of estimating the contribution of PV power to the grid without knowing the actual production and consumption “behind-the-meter”. The main approach using a physical model was compared with persistence (forecast with yesterday’s values) and black box models represented by linear regression and an artificial neural network. The artificial neural network performed best with a normalized (with installed power) RMSE of about 10 %. Linear regression was only slightly inferior while the physical model performed worse (about 15 %). The physical model only predicts the gross production and needs information about the on site consumption in order to predict the net load. Here the consumption was predicted using persistence. This assumption did not hold up for the physical model to match the performance of the black box models. Thus, a better description of the gross consumption is needed in order to make the physical model more competitive.