An Approach to Forecasting Short-Term Photovoltaic Power Generation

As photovoltaic power generation technology matures, an increasing number of distributed photovoltaics are being integrated into the distribution network. Due to the inherent randomness and intermittency of their output, this integration poses a series of challenges for grid-connected distribution systems. One effective approach to mitigate these adverse effects is through accurate forecasting of photovoltaic power generation, which can alleviate impacts on the grid and enhance power quality. This paper proposes a PV power prediction model that leverages the K-Means ++ algorithm, an improved similar day method, and Gated Recurrent Unit (GRU) networks.