Short-term power forecasting for photovoltaic generation based on wavelet neural network and residual correction of Markov chain

With large-scale photovoltaic power generation system implementing grid-connected operation, it is essential for the accurate and reliable power forecasting of the photovoltaic generation to reduce the impact of uncertainty on the power network. A method of power forecasting of the photovoltaic generation based on wavelet neural network and residual correction of Markov chain is proposed in this paper. Firstly the various meteorological factors and the correlation coefficient are analyzed to identify the key meteorological factors of the photovoltaic generation. Then a wavelet neural network prediction model is established to forecast the power output of the photovoltaic generation. Finally the forecasting power of the photovoltaic generation can be modified with the residual correction of Markov chain. The case at an area in Beijing is used to verify the applicability and high accuracy of the proposed method.