Building electrical consumption patterns forecasting based on a novel hybrid deep learning model
Accurate prediction of electrical energy consumption in smart buildings is a critical challenge for optimizing energy management systems, reducing costs, and improving overall efficiency. Existing models often fail to account for the complex and nonlinear characteristics of energy consumption patter...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-06-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025005997 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850028655180251136 |
|---|---|
| author | Nasser Shahsavari-Pour Azim Heydari Farshid Keynia Afef Fekih Aylar Shahsavari-Pour |
| author_facet | Nasser Shahsavari-Pour Azim Heydari Farshid Keynia Afef Fekih Aylar Shahsavari-Pour |
| author_sort | Nasser Shahsavari-Pour |
| collection | DOAJ |
| description | Accurate prediction of electrical energy consumption in smart buildings is a critical challenge for optimizing energy management systems, reducing costs, and improving overall efficiency. Existing models often fail to account for the complex and nonlinear characteristics of energy consumption patterns, resulting in suboptimal prediction performance. This paper addresses the problem of accurate energy forecasting by proposing an intelligent hybrid model that integrates advanced feature selection, signal decomposition, and deep learning techniques. Specifically, the proposed model comprises three key components: (i) a mutual information-based feature selection method to identify the most significant input variables influencing energy consumption; (ii) a variational mode decomposition (VMD) approach to decompose the original energy consumption signal into intrinsic mode functions (IMFs), capturing relevant trends and eliminating noise; and (iii) a long short-term memory (LSTM) neural network to perform time-series forecasting of the target energy consumption values. The performance of the proposed model was evaluated using real-world datasets collected from a smart two-story residential building in Houston, Texas, USA. A comparative analysis was conducted against benchmark models, including the generalized regression neural network (GRNN) and adaptive neuro-fuzzy inference system (ANFIS), to validate the efficacy of the proposed approach. The results demonstrate that the proposed hybrid model achieves a significantly lower average root mean square error (RMSE) of 0.1192, compared to 0.264 for the GRNN and 0.319 for the ANFIS models, indicating superior prediction accuracy. These findings highlight the effectiveness of integrating mutual information, VMD, and LSTM for improving energy consumption forecasting in smart buildings. The proposed model provides a robust and accurate tool for energy management, enabling smart buildings to enhance operational optimization and energy efficiency. |
| format | Article |
| id | doaj-art-4fabedfbdba64cd78d8be507585fec18 |
| institution | DOAJ |
| issn | 2590-1230 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj-art-4fabedfbdba64cd78d8be507585fec182025-08-20T02:59:46ZengElsevierResults in Engineering2590-12302025-06-012610452210.1016/j.rineng.2025.104522Building electrical consumption patterns forecasting based on a novel hybrid deep learning modelNasser Shahsavari-Pour0Azim Heydari1Farshid Keynia2Afef Fekih3Aylar Shahsavari-Pour4Department of Industrial Engineering, Vali-e-Asr University of Rafsanjan, IranDepartment of Energy, Institute of Science and High Technology and Environmental Sciences Graduate University of Advanced Technology, Kerman, Iran; Corresponding author.Department of Energy Management and Optimization, Institute of Science and High Technology and Environmental Sciences Graduate University of Advanced Technology, Kerman, IranDepartment of Electrical and Computer Engineering, University of Louisiana at Lafayette, Lafayette, LA, 70504, USADepartment of Mathematical Sciences, Sharif University of Technology, Tehran, IranAccurate prediction of electrical energy consumption in smart buildings is a critical challenge for optimizing energy management systems, reducing costs, and improving overall efficiency. Existing models often fail to account for the complex and nonlinear characteristics of energy consumption patterns, resulting in suboptimal prediction performance. This paper addresses the problem of accurate energy forecasting by proposing an intelligent hybrid model that integrates advanced feature selection, signal decomposition, and deep learning techniques. Specifically, the proposed model comprises three key components: (i) a mutual information-based feature selection method to identify the most significant input variables influencing energy consumption; (ii) a variational mode decomposition (VMD) approach to decompose the original energy consumption signal into intrinsic mode functions (IMFs), capturing relevant trends and eliminating noise; and (iii) a long short-term memory (LSTM) neural network to perform time-series forecasting of the target energy consumption values. The performance of the proposed model was evaluated using real-world datasets collected from a smart two-story residential building in Houston, Texas, USA. A comparative analysis was conducted against benchmark models, including the generalized regression neural network (GRNN) and adaptive neuro-fuzzy inference system (ANFIS), to validate the efficacy of the proposed approach. The results demonstrate that the proposed hybrid model achieves a significantly lower average root mean square error (RMSE) of 0.1192, compared to 0.264 for the GRNN and 0.319 for the ANFIS models, indicating superior prediction accuracy. These findings highlight the effectiveness of integrating mutual information, VMD, and LSTM for improving energy consumption forecasting in smart buildings. The proposed model provides a robust and accurate tool for energy management, enabling smart buildings to enhance operational optimization and energy efficiency.http://www.sciencedirect.com/science/article/pii/S2590123025005997Building energy consumption patternLoad forecastingSmart buildingDeep learning modelsEnergy management |
| spellingShingle | Nasser Shahsavari-Pour Azim Heydari Farshid Keynia Afef Fekih Aylar Shahsavari-Pour Building electrical consumption patterns forecasting based on a novel hybrid deep learning model Results in Engineering Building energy consumption pattern Load forecasting Smart building Deep learning models Energy management |
| title | Building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| title_full | Building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| title_fullStr | Building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| title_full_unstemmed | Building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| title_short | Building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| title_sort | building electrical consumption patterns forecasting based on a novel hybrid deep learning model |
| topic | Building energy consumption pattern Load forecasting Smart building Deep learning models Energy management |
| url | http://www.sciencedirect.com/science/article/pii/S2590123025005997 |
| work_keys_str_mv | AT nassershahsavaripour buildingelectricalconsumptionpatternsforecastingbasedonanovelhybriddeeplearningmodel AT azimheydari buildingelectricalconsumptionpatternsforecastingbasedonanovelhybriddeeplearningmodel AT farshidkeynia buildingelectricalconsumptionpatternsforecastingbasedonanovelhybriddeeplearningmodel AT afeffekih buildingelectricalconsumptionpatternsforecastingbasedonanovelhybriddeeplearningmodel AT aylarshahsavaripour buildingelectricalconsumptionpatternsforecastingbasedonanovelhybriddeeplearningmodel |