Research on Pearson correlation and improved CNN-LSTM algorithm for predicting photovoltaic power generation

To enhance the grid connection reliability of distributed photovoltaic power plants, a hybrid prediction model integrating a convolutional neural network and short-term memory is proposed. Spatial features are extracted via CNN, while LSTM captures temporal dependencies in power generation data. Initially, leveraging clustering and partitioning of weather data, the Pearson correlation coefficient method is employed to analyze correlations between meteorological factors (e.g., solar radiation, temperature, relative humidity) and photovoltaic power generation. Subsequently, the Sparrow Search Algorithm is applied to optimize the prediction model. Experimental findings using photovoltaic power generation data in Dingbian County reveal that the Sparrow Optimization Algorithm significantly enhances prediction accuracy and improves the scheduling stability of distributed photovoltaic power stations.