Study on Short-Term Photovoltaic Power Prediction Based on SCSSA-CNN-BiLSTM

To enhance the accuracy of photovoltaic power generation, this paper integrates the sparrow search optimization algorithm (SCSSA) with convolutional neural networks (CNN) and bidirectional long short-term memory neural networks (BiLSTM). This integration allows the feature data from the CNN to be incorporated into the BiLSTM, while the SCSSA, based on the joint positive cosine and cauchy theorem, creates a new hybrid neural network model for predicting photovoltaic power. Within this model, the BiLSTM establishes the temporal relationships among historical data, leading to improved photovoltaic power prediction outcomes. Ultimately, the SCSSA-CNN-BiLSTM model demonstrates exceptional stability and accuracy in forecasting photovoltaic power, meeting the current accuracy requirements for short-term solar power generation.