Short-term PV-Wind forecasting of large-scale regional site clusters based on FCM clustering and hybrid Inception-ResNet embedded with Informer

Abstract

In order to cope with the challenge that the high proportion of new energy generation for the stable operation of the power grid, this paper proposes an innovative short-term power forecasting model for regional site clusters based on fuzzy c-means (FCM) clustering and hybrid Inception-ResNet deep neural network embedded with Informer. Firstly, multiple wind farms and photovoltaic sites are clustered into different groups for popular clustering prediction owing to FCM clustering algorithm. Secondly, numerous strong factors are selected based on the combination of the linear and nonlinear correlation analysis between the variables and power generation. Furthermore, the improved gray wolf algorithm (GWO) can determine the optimal parameters of deep network model and the Informer and Inception are integrated which is fairly advanced to capture temporal relationship and potent feature extraction. Finally, the wind and photovoltaic dataset in western China is employed to verify our model and the results demonstrate that ours outperforms other algorithms with 5.400% and 4.200% higher R2 and 2.525% and 2.090% lower MAPE in the wind and solar forecasting, which simultaneously improves the accuracy and efficiency of prediction.