Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD

Accurate long-term power load forecasting in the grid is crucial for supply–demand balance analysis in new power systems. It helps to identify potential power market risks and uncertainties in advance, thereby enhancing the stability and efficiency of power systems. Given the temporal and nonlinear...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiaoyu Liu, Jiangfeng Song, Hai Tao, Peng Wang, Haihua Mo, Wenjie Du
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Energies
Subjects:
Online Access:https://www.mdpi.com/1996-1073/18/11/2675
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849722304700874752
author Xiaoyu Liu
Jiangfeng Song
Hai Tao
Peng Wang
Haihua Mo
Wenjie Du
author_facet Xiaoyu Liu
Jiangfeng Song
Hai Tao
Peng Wang
Haihua Mo
Wenjie Du
author_sort Xiaoyu Liu
collection DOAJ
description Accurate long-term power load forecasting in the grid is crucial for supply–demand balance analysis in new power systems. It helps to identify potential power market risks and uncertainties in advance, thereby enhancing the stability and efficiency of power systems. Given the temporal and nonlinear features of power load, this paper proposes a hybrid load-forecasting model using attention mechanisms, CNN, and BiLSTM. Historical load data are processed via CEEMDAN, K-means clustering, and VMD for significant regularity and uncertainty feature extraction. The CNN layer extracts features from climate and date inputs, while BiLSTM captures short- and long-term dependencies from both forward and backward directions. Attention mechanisms enhance key information. This approach is applied for seasonal load forecasting. Several comparative experiments show the proposed model’s high accuracy, with MAPE values of 1.41%, 1.25%, 1.08% and 1.67% for the four seasons. It outperforms other methods, with improvements of 0.25–2.53 GWh<sup>2</sup> in MSE, 0.15–0.1 GWh in RMSE, 0.1–0.74 GWh in MAE and 0.22–1.40% in MAPE. Furthermore, the effectiveness of the data processing method and the impact of training data volume on forecasting accuracy are analyzed. The results indicate that decomposing and clustering historical load data, along with large-scale data training, can both boost forecasting accuracy.
format Article
id doaj-art-5d89c4cee1cb456694d4d94b6972e7da
institution DOAJ
issn 1996-1073
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Energies
spelling doaj-art-5d89c4cee1cb456694d4d94b6972e7da2025-08-20T03:11:22ZengMDPI AGEnergies1996-10732025-05-011811267510.3390/en18112675Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMDXiaoyu Liu0Jiangfeng Song1Hai Tao2Peng Wang3Haihua Mo4Wenjie Du5Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaGuangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaGuangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaGuangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaGuangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaGuangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, ChinaAccurate long-term power load forecasting in the grid is crucial for supply–demand balance analysis in new power systems. It helps to identify potential power market risks and uncertainties in advance, thereby enhancing the stability and efficiency of power systems. Given the temporal and nonlinear features of power load, this paper proposes a hybrid load-forecasting model using attention mechanisms, CNN, and BiLSTM. Historical load data are processed via CEEMDAN, K-means clustering, and VMD for significant regularity and uncertainty feature extraction. The CNN layer extracts features from climate and date inputs, while BiLSTM captures short- and long-term dependencies from both forward and backward directions. Attention mechanisms enhance key information. This approach is applied for seasonal load forecasting. Several comparative experiments show the proposed model’s high accuracy, with MAPE values of 1.41%, 1.25%, 1.08% and 1.67% for the four seasons. It outperforms other methods, with improvements of 0.25–2.53 GWh<sup>2</sup> in MSE, 0.15–0.1 GWh in RMSE, 0.1–0.74 GWh in MAE and 0.22–1.40% in MAPE. Furthermore, the effectiveness of the data processing method and the impact of training data volume on forecasting accuracy are analyzed. The results indicate that decomposing and clustering historical load data, along with large-scale data training, can both boost forecasting accuracy.https://www.mdpi.com/1996-1073/18/11/2675long-term load forecastingCNN-BiLSTMattention mechanismCEEMDANK-means clusteringVMD
spellingShingle Xiaoyu Liu
Jiangfeng Song
Hai Tao
Peng Wang
Haihua Mo
Wenjie Du
Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
Energies
long-term load forecasting
CNN-BiLSTM
attention mechanism
CEEMDAN
K-means clustering
VMD
title Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
title_full Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
title_fullStr Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
title_full_unstemmed Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
title_short Quarter-Hourly Power Load Forecasting Based on a Hybrid CNN-BiLSTM-Attention Model with CEEMDAN, K-Means, and VMD
title_sort quarter hourly power load forecasting based on a hybrid cnn bilstm attention model with ceemdan k means and vmd
topic long-term load forecasting
CNN-BiLSTM
attention mechanism
CEEMDAN
K-means clustering
VMD
url https://www.mdpi.com/1996-1073/18/11/2675
work_keys_str_mv AT xiaoyuliu quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd
AT jiangfengsong quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd
AT haitao quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd
AT pengwang quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd
AT haihuamo quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd
AT wenjiedu quarterhourlypowerloadforecastingbasedonahybridcnnbilstmattentionmodelwithceemdankmeansandvmd