KETEP 세계에너지시장정보

1. An optimization method for cross-site photovoltaic power generation power prediction models based on dual-driven AI for knowledge and data, characterized by comprising the following steps:Based on source and target site data, determine a photovoltaic power prediction model that fuses physics and data;Performing hyperparameter tests on at least one source station based on the source station data and the photovoltaic power prediction model to obtain historical optimization process data;Extracting meta-features from the target station data, and retrieving similar source sites from the historical optimization process data based on the meta-features, generating an initial set of hyperparameter combinations for the target sites;Extracting learning rate process knowledge from the historical optimization process data corresponding to the initial hyperparameter combination set, predicting the optimal peak learning rate and stable critical learning rate at the target site based on the learning rate process knowledge, and training the dynamic optimization learning rate at the target site based on the optimal peak learning rate and stable critical learning rate;Based on the optimized learning rate, cost-sensitive objective function, and target site data, Bayesian optimization search is used to obtain the optimal hyperparameters of the target site and the corresponding photovoltaic power prediction model;Based on the learning rate process knowledge, the optimal peak learning rate and stable critical learning rate at the target site are predicted, and the learning rate is dynamically optimized by training at the target site based on the optimal peak learning rate and the stable critical learning rate. The steps include:Predicting the optimal peak learning rate and stable critical learning rate of the target site based on the meta-features and model scale features of the target site;Based on the similarity weights between the similar source site and the target site, the optimal peak learning rate and stable critical learning rate of the learning rate process knowledge of the similar source site are weighted respectively to obtain the prior optimal peak learning rate and prior stable critical learning rate;The predicted optimal peak learning rate is fused with the prior optimal peak learning rate, and the predicted stable critical learning rate with the prior stable critical learning rate is fused, and dynamic learning rate training is performed separately after fusion.