The ultra-short-term prediction of photovoltaic power system is of great significance to the safe operation of power system. However, the daily and annual periodicity of solar radiation brings strong non-stationary to the photovoltaic power series, which makes it difficult to predict the photovoltaic power series. In order to overcome the problem that the detailed clear sky model relies on a large number of plant parameters, this paper proposes a prediction algorithm based on a modified clear sky model. Firstly, historical data and geographical location of a certain site are utilized to build a modified clear sky model, then the theoretical clear sky power curve fitted by the modified clear sky model are divided by the actual historical power output to get a stabilized time series, finally the method of "online update" is adopted to forecast PV plant power output in the ultra-short-term time horizon (0-4 h). The test results of a photovoltaic power station in Ningxia show that the proposed model can reduce the prediction error of the 4th hour to about 3%.
[1]
Nicholas A. Engerer,et al.
KPV: A clear-sky index for photovoltaics
,
2014
.
[2]
J. F. Randall,et al.
Is AM1.5 applicable in practice? Modelling eight photovoltaic materials with respect to light intensity and two spectra
,
2003
.
[3]
T. Funabashi,et al.
Application of Recurrent Neural Network to Short-Term-Ahead Generating Power Forecasting for Photovoltaic System
,
2007,
2007 IEEE Power Engineering Society General Meeting.
[4]
H. Pedro,et al.
Assessment of forecasting techniques for solar power production with no exogenous inputs
,
2012
.
[5]
Nicholas A. Engerer,et al.
Validating nine clear sky radiation models in Australia
,
2015
.
[6]
Henrik Madsen,et al.
Online short-term solar power forecasting
,
2009
.