Research on Short-term Prediction of Photothermal Power Generation Based on FCM-LSTM

With the proposal of “Double Carbon” goal, clean energy such as solar energy has been vigorously developed in China. Short-term prediction of the power generation of concentrated solar power can effectively deal with the impact of randomness and volatility of solar energy, and make preparations for power grid dispatching. In this paper, taking a concentrated solar power station in Qinghai as an example, the fuzzy C-means clustering algorithm is used to classify the experimental data, and then the meteorological factors are weighted by analyzing the correlation degree between different types of meteorological factors and power output. Finally, the long short-term memory neural network prediction models under different clusters are constructed. The results show that, compared with the unweighted model and the traditional long short-term memory neural network model, the weighted long short-term memory prediction model based on fuzzy C-means clustering has a good effect, greatly reduces the standard deviation, and verifies the effectiveness of the prediction model.