Short-Term Photovoltaic Output Forecasting with Weakly Related Meteorological Data

DOI

10.2991/cisia-15.2015.115How to use a DOI?

Keywords

distributed photovoltaic; PV output forecasting; support vector machine; parameter selected

Abstract

Photovoltaic (PV) output is influenced by many meteorological factors. The significant degree of meteorological data influences the accuracy of forecasting result. This paper proposed a short-term PV output forecasting method while the weather data and PV output data were weak correlated. By analyzing meteorological historical data and PV output historical data, the main factors effecting PV generation were found out by Pearson correlation coefficient. Based on relevant factors, fuzzy clustering analysis method was used to select similar days, and then support vector regression (SVR) forecasting model was built. SVR model has excellent learning ability for small sample. To determine model parameters, a two-step method was proposed. First, using the global search method to determine the value of parameter ? and the appropriate range of kernel parameter p and regularization parameter C, then using self-adaptive differential evolution algorithm to find the optimal p and C, in order to improve the forecast accuracy when parameter ? was selected in large scale. Examples show that the method proposed in this paper has good forecasting ability when the weather data and PV output data are weak correlated.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - C.H Lin
AU  - Y. Xiao
AU  - J.G Chen
AU  - X.K Wen
AU  - C. Du
PY  - 2015/06
DA  - 2015/06
TI  - Short-Term Photovoltaic Output Forecasting with Weakly Related Meteorological Data
BT  - Proceedings of the International Conference on Computer Information Systems and Industrial Applications
PB  - Atlantis Press
SP  - 426
EP  - 429
SN  - 2352-538X
UR  - https://doi.org/10.2991/cisia-15.2015.115
DO  - 10.2991/cisia-15.2015.115
ID  - Lin2015/06
ER  -