Photovoltaic power generation prediction based on k-means clustering analysis and GRO-CNN-LSTM Attention

In order to improve the accuracy of photovoltaic power prediction, this paper proposes a GRO-CNN-LSTM Attention prediction model, which improves the prediction accuracy by introducing the Gold Rushing Optimization Algorithm (GRO) and SE attention mechanism. Secondly, the photovoltaic data is divided into three categories based on weather types using the k-means clustering analysis algorithm: sunny, cloudy, and rainy. Finally, photovoltaic power generation predictions were conducted under three different weather types, and the proposed prediction model was compared and analyzed with other prediction models. The results showed that the CNN-LSTM prediction model, which introduced GRO and SE attention mechanism optimization, had a decrease in MAE, RMSE, and MAPE compared to the original model under three different weather types.