A load fluctuation prediction method for high-proportion photovoltaic resources

With the rapid access of renewable energy solar photovoltaic panels to the power grid system, the accuracy of photovoltaic power generation prediction plays an important role in improving system reliability and maintaining power grid stability. This paper introduces the whale optimization algorithm (WOA) for adjustment. By combining the convolutional neural network (CNN) and the bidirectional long short-term memory network (BiLSTM), supplemented by the attention mechanism (Attention) model, a new VMD-WOA-BiLSTM-Attention(VWBA) photovoltaic power generation prediction model is constructed. The WOA-optimized CNN-BiLSTM network can better summarize and extract long-term correlation features in time series, while using the attention mechanism to focus on important information. The model first uses the variational mode decomposition (VMD) method to process the original power load data to extract more representative features and reduce noise interference. Then the WOA algorithm is used to optimize the three hyperparameters of CNN-BiLSTM(CB): learning rate, number of hidden layer nodes, and regularization coefficient, which further improves the performance of the model. In addition, this paper also introduces the attention mechanism to focus on the dynamic changes of the load at critical moments. Experimental results show that the VWBA network achieves good performance in the short-term photovoltaic power prediction task.