Abstract：In order to improve the accuracy of photovoltaic power generation prediction,a new BP neutral network power forecasting model is developed based on the principle of similar day and Markov chain. Using the similar day data as the input of the forecast model,the error correction date can be obtained based on the Markov chain model,and the preliminary forecast data can be attained through BP neural network training.According to the error correction date the new forecast date can be gotten. At last,by comparing the results obtained by traditional algorithm,it is shown that the improved algorithm of prediction accuracy is higher than that of traditional algorithm,verifying the effectiveness of the proposed model.

Keyword：photovoltaic generation;similar day algorithm;BP neutral network;Markov chain;