Short-Term Forecasting Error Assessment of Solar Power Plant Generation and the Error Influence on Plant Economics in Conditions in Russia

• Published in: Applied solar energy 2021, Vol.57 (4), p.347-353
• Main Authors: Kiseleva, S. V., Lisitskaya, N. V., Mordynskiy, A. V., Frid, S. E.
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Economic analysis; Economic forecasting; Economic impact; Economics; Electrical Machines and Networks; Engineering; Error analysis; Mathematical models; Neural networks; Numerical prediction; Numerical weather forecasting; Photovoltaics; Power Electronics; Power plants; Predicting Solar Radiation; Root-mean-square errors; Simulation; Solar energy; Solar power plants; Statistical analysis; Statistical models; Weather forecasting
• ISSN: 0003-701X; 1934-9424
• DOI: 10.3103/S0003701X2104006X

The production forecast error for an experimental photovoltaic installation in Moscow and an autonomous solar power plant (SPP) in Yailyu village, Altai Republic, has been estimated by dynamic simulation of SPP operation. It is shown that the root-mean-square error of the production forecast according to numerical weather prediction data (NWP, ICON model) for the considered installations is 16–24% (normalized to the installed capacity of the SPP). A weak and spatially nonuniform seasonal dependence of the error has been found. A statistical forecasting method using artificial neural networks (ANNs) was tested the effectiveness of which was compared with dynamic simulation based on ICON data. Statistical models showed more accurate results of production forecast for the Moscow region, while for the Yailyu SPP the situation is the opposite: dynamic simulation using NWP showed higher efficiency in terms of the root-mean-square error of the forecast. The analysis of the economic effect of errors shows that for an error of 13–14%, the economic losses of the SPP from not complying with the dispatch schedule are 15–20%.