Prediction Method of Photovoltaic Active Power Based on Clustering and Variational Mode Decomposition- Gated Recurrent Unit

• Published in: 2024 3rd International Conference on Energy, Power and Electrical Technology (ICEPET) pp. 162 - 166
• Main Authors: Hu, Hongtao, Xie, Weirong, Li, Bing, Han, Libao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.05.2024
• Subjects: Accuracy; Computational modeling; Distributed databases; Gated Recurrent Unit; K-means; Logic gates; Photovoltaic systems; power prediction; Prediction algorithms; Predictive models; Variational Mode Decomposition
• DOI: 10.1109/ICEPET61938.2024.10626523

In the photovoltaic power generation process, the AC active power of the inverter is the effective power actually output to the grid, which has the characteristics of strong volatility and instability, and its prediction depends very much on meteorological data such as irradiance, sunshine time, ambient temperature and so on. Due to the geographical location of the distributed photovoltaic power station and construction costs and other reasons, it is difficult to obtain detailed meteorological data, so it is difficult to achieve accurate prediction of the power. To solve the above problems, this paper proposes a method of photovoltaic active power prediction based on clustering and Variational Mode Decomposition-Gated Recurrent Unit. Firstly, K-means algorithm is used to cluster the power, and the influence of different weather on the power is considered. Then, the power after clustering is decomposed by Variational Mode Decomposition algorithm to fully extract useful information in the historical power, and the results of Variational Mode Decomposition are added to the original data set as new features. Finally, the data set is fed into the Gated Recurrent Unit network for training and prediction. The experimental results are verified by actual data and compared with Long Short-Term Memory, Recurrent Neural Network and Gated Recurrent Unit models. The results show that this method can effectively improve the accuracy of active power prediction of distributed photovoltaic in the absence of meteorological data.