Photovoltaic Power Prediction Model Based on VMD-IRIME-LSTM

Accurate prediction of photovoltaic power generation directly affects the stability of the power grid. By optimizing the number of hidden layer units, the maximum number of iterations and the initial learning rate threshold of the long-term and short-term memory network (LSTM), the photovoltaic power prediction can be improved. Therefore, this paper proposes an improved rime optimization algorithm (IRIME) with cube chaotic mapping, butterfly optimization algorithm and triangle walk strategy to optimize the combined prediction model based on LSTM and variable mode decomposition (VMD). Firstly, the optimal decomposition K value is found by the sparse index adaptive method and the original data is decomposed by VMD. Secondly, IRIME algorithm is introduced to optimize the three core parameters of LSTM network. Finally, each decomposition sequence is independently predicted by LSTM and the prediction results are obtained by superposition. Through the simulation analysis of measured photovoltaic data, the root mean square error of the model is reduced by 37.1%, 36.7% and 20.2% respectively compared with VMD-RIME-LSTM model in sunny, abrupt and cloudy days. The prediction accuracy is also higher than that of the other two combined prediction models, indicating that the model successfully improves the prediction accuracy of photovoltaic power generation.