Short Term Photovoltaic Power Combination Prediction Method Based on Similar Day Selection and Data Reconstruction
A short-term photovoltaic power combination prediction method based on similar day selection and data reconstruction is proposed to address the strong randomness of photovoltaic power. Firstly, clustering analysis of photovoltaic power is performed using the kernel fuzzy C-means algorithm, and the m...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
State Grid Energy Research Institute
2024-12-01
|
| Series: | Zhongguo dianli |
| Subjects: | |
| Online Access: | https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202406005 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | A short-term photovoltaic power combination prediction method based on similar day selection and data reconstruction is proposed to address the strong randomness of photovoltaic power. Firstly, clustering analysis of photovoltaic power is performed using the kernel fuzzy C-means algorithm, and the main influencing features are extracted through the maximum information coefficient. Secondly, the cooperative game theory is used to calculate the comprehensive correlation coefficient between the predicted days and the historical days, and the historical days with strong correlation are selected to construct the training set. Then, the variational mode decomposition method is used to decompose the photovoltaic power into several subsequences, and the permutation entropy is calculated and reconstructed into trend, low-frequency, and high-frequency terms. Finally, the long short-term memory neural networks are used to predict the trend and low-frequency items, while the convolutional neural network-bidirectional long short-term memory-attention models are used to predict the high-frequency items. The final prediction result is obtained by overlaying the results. Through practical examples, it has been verified that under different weather conditions, the overall prediction error of the model is the smallest, which can effectively improve the prediction accuracy. |
|---|---|
| ISSN: | 1004-9649 |