Improving the accuracy of photovoltaic power generation prediction has important significance for ensuring the security and stability of power system and improving the development and utilization of solar energy resources. A short-term power forecasting model for photovoltaic power generation is proposed in this paper. This model is based on weather similiarty degree and improved cuckoo search algorithm(ICS),which is used to optimize Elman neural network. Firstly,in order to select the similar days with higher similarity,a similarity calculation method based on distance and angle trend is proposed. Secondly,the improved cuckoo algorithm is used to optimize the weight and threshold of Elman neural network,and a short-term power prediction model for photovoltaic power generation is established. Finally,the prediction results of the proposed model is compared with the results of traditional model. The results show that the prediction accuracy of this method is higher.