Research on Photovoltaic Power Prediction Based on Multi-model Fusion

Accurately predicting photovoltaic power generation has a pivotal role in ensuring the safe and stable operation of the power system. This manuscript proposes a multi-network fusion short-term PV power forecasting model. Firstly, the PV power was decomposed into steady-state and unsteady components by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and the unsteady components were decomposed twice by variational mode decomposition (VMD). Secondly, the Catboost network is employed to predict steady-state decomposition power, and the WTCN-BiGRU-Attention network optimized by IWOA is employed to predict unsteady component PV power. Finally, the short-term prediction of PV power is realized by integrating the prediction results of each model component. A photovoltaic power plant’s measured power generation for the whole year of 2018 is utilized to validate the prediction model, and the results show that the proposed multi-model fusion network with quadratic mode decomposition has a high accuracy for power generation prediction.