A Short-Term Photovoltaic Power Prediction Model Based on Cnn-Gru with Moss Growth Optimization and Attention Mechanism

The output power of photovoltaic (PV) power plants is subject to fluctuations due to meteorological conditions and defects in power equipment. Therefore, accurate photovoltaic power prediction is of great importance for grid scheduling. In order to address the issue of the limited accuracy of a single gated recurrent unit model (GRU) when applied to photovoltaic power prediction, A short-term photovoltaic power prediction model based on CNN-GRU with moss growth optimization and attention mechanism has been developed. The thresholds and parameters of the GRU and the attention model are optimized by the moss growth optimization to establish a short-term photovoltaic power prediction model with near-optimal parameters. Test results based on a photovoltaic plant dataset show that the proposed model in this paper has a low error rate, a higher fitting coefficient, and enhanced stability in both cloudy and sunny weather environments. The model exhibited a high degree of fit, reaching 0.9923 for sunny days and 0.9661 for cloudy days, which substantiates the efficacy of the proposed approach.