Load data recovery method based on SOM-LSTM neural network
In the collection and transmission of power big data, the problem of data missing exists. In response to this problem, this paper proposes a power data detection and repair method based on SOM-LSTM. Firstly, a large amount of collected power data is analyzed and the type of missing data is determine...
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | article |
| Langue: | EN |
| Publié: |
Elsevier
2022
|
| Sujets: | |
| Accès en ligne: | https://doaj.org/article/141c36bd4c034a11bfb0fc0ec12f8f19 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| Résumé: | In the collection and transmission of power big data, the problem of data missing exists. In response to this problem, this paper proposes a power data detection and repair method based on SOM-LSTM. Firstly, a large amount of collected power data is analyzed and the type of missing data is determined. Then, the SOM is used to classify the power data. The LSTM is trained according to the characteristic values of different users to complete the detection and repair of different types of missing power data. Finally, the analysis is based on actual data in some regional loads of China. Experimental results show that, compared with the extreme learning machine (ELM) and LSTM, the proposed SOM-LSTM model reduces the mean absolute error (MAE) by 0.2498 and 0.3425, and the root mean square error (RMSE) by 0.1048 and 0.1469, respectively. |
|---|