Deep Learning Model for Short-term Photovoltaic Power Prediction Based on Data Augmentation and Quantum Computing

With the growth of renewable energy demand, accurate prediction of short-term photovoltaic (PV) power is crucial for grid management and energy dispatch. In this study, a short-term PV power prediction method based on the VAE-GWO-VQC-GRU hybrid model is proposed. First, data augmentation of PV power generation data using Variational Auto-Encoder (VAE) addresses the model training difficulties due to insufficient data volume. Second, prediction models combining Variational Quantum Circuits (VQC) and Gated Recurrent Units (GRU) are constructed for data under three different weather conditions, respectively. In these models, VQC extracts nonlinear dependencies, while GRU refines temporal features. Finally, the model hyperparameters are optimized using the Grey Wolf Optimizer (GWO) algorithm to further improve prediction accuracy. Experiments show that the hybrid model achieves significant accuracy and applicability in short-term PV power generation prediction, particularly excelling in coping with weather variations compared to traditional methods.