Short term photovoltaic power prediction based on FWA-CNN-GRU neural network. Issue 1 (1st July 2022)

• Record Type: Journal Article
• Authors: Wang, Xiaohuan; Zhao, Quanming; Li, Ke; Yang, Tianyi; Fan, Chuang
• Is Part Of: Journal of physics. Volume 2303:Issue 1(2022)
• Publication Date: 2022-07-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2303/1/012024
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: In order to achieve accurate prediction of short-term photovoltaic power generation, this paper establishes a photovoltaic power prediction model based on FWA-CNN-GRU through the combination and optimization of neural network. Firstly, this paper crunches data and establishes the test sets and training sets. Then, the prediction model based on CNN-GRU is built through Convolutional Neural Networks (CNN) and Gated Recurrent Unit (GRU) neural network. Meanwhile, the Fireworks Algorithm (FWA) is introduced to search the optimal value of the super parameters of the network model. A prediction model that is based on FWA-CNN-GRU is established. Finally, the prediction is made according to different seasons. The experimental comparison shows that the average Root Mean Square Error and Mean Absolute Error of the optimized model are reduced by 9.190% and 6.095% respectively, and the average coefficient of determination reaches 96.83%. It shows that FWA-CNN-GRU model has higher prediction accuracy and can provide a theoretical basis for the operation and regulation of power grid.