Article Details

Title · Authors · Abstract · Keywords · DOI · References

Research on Building Energy Consumption Prediction Method Based on Hybrid Neural Network

TONG Yuxing, XIONG Yi

(Guangdong Provincial Railway Planning and Design Institute Co., Ltd., Guangzhou 510699, China)

Section：Intelligent Manufacturing · Vol. 2026, Issue 1 · Pages：164-168 · 2096-4706（2026）01-0164-05 · CLC number：TP391 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2026.01.031

Abstract

Accurate prediction of building energy consumption is crucial for energy management and policy-making. However, due to the complex interaction of multiple factors, prediction faces many challenges. This paper proposes a hybrid prediction model (KCL) integrating Kalman Filter, Continuous Wavelet Transform (CWT), and Long Short-Term Memory (LSTM) neural network. This model combines the advantages of each technology in denoising, time-frequency analysis, and time series modeling. Experiments based on public building electricity consumption datasets show that the KCL model significantly outperforms the traditional LSTM and other wavelet-based models in multiple evaluation metrics. The prediction accuracy is significantly improved, which verifies the effectiveness and robustness of the model in building energy consumption prediction.

Keywords

building energy consumption; Neural Network; LSTM; intelligent building

References

  1. [1] Jiang Yizhou, Ma Lin, Wei Cun, etc Analysis of Urban Building Energy Consumption and Carbon Emissions Based on Time Series [J]. Building Science, 2025, 41 (4): 66-75
  2. [2] Jiang Xinbo, Zhang Yushuai, Liu Yuxin, etc Carbon emissions and cost optimization design of building lifecycle under carbon pricing mechanism [J]. Journal of Xi'an University of Architecture and Technology: Natural Science Edition, 2025, 57 (1): 99-106
  3. [3] Qu Minglu, Du Shanghe, Zhang Xinlin, etc Research on Energy Consumption Prediction of Building Heating System Based on Model Identification Method [J]. Journal of Refrigeration, 2025, 46 (3): 145-150166
  4. [4] Sun Yaolong Exploration of Energy saving Renovation of Existing Buildings under the Concept of Green Building [J]. Engineering Seismic and Reinforcement Renovation, 2024, 46 (6): 201-202
  5. [5] Zhu Guoqing, Liu Xiancheng, Tian Congxiang Short term energy consumption prediction model for public buildings based on long short-term memory networks [J]. Journal of Jilin University: Engineering Edition, 2024, 54 (7): 2009-2014
  6. [6] SADOWSKY J. Investigation of Signal Characteristics Using the Continuous Wavelet Transform [J].Johns Hopkins Apl Technical Digest，1996，17（3）：258-269.
  7. [7] KIM D S，LEE Y J，CHIN K，et al. Implementation of a Long Short-Term Memory Transfer Learning (LSTM-TL)-Based Data-Driven Model for Building Energy Demand Forecasting [J/OL].Sustainability，2023，15（3）：2340[2025-06-28].https://www.mdpi.com/2071-1050/15/3/2340.
  8. [8] ZOU X L，WAGNER N，SUNDHARARAJAN. Analysis of Consumer Online Resale Behavior Measurement Based on Machine Learning and BP Neural Network [J].Journal of Intelligent & Fuzzy Systems，2021，40（2）：2121-2132.
  9. [9] ZHANG Y L，DU D S，SHI S，et al. Effects of the Earthquake Nonstationary Characteristics on the Structural Dynamic Response: Base on the BP Neural Networks Modified by the Genetic Algorithm [J/OL].Buildings，2021，11（2）：69[2025-06-28].https://www.mdpi.com/2075-5309/11/2/69.
  10. [10] ZHANG Y M，YANG J，ZHOU S，et al. Building Energy Consumption Prediction Based on Bayesian Optimized LSTM model [C]//International Conference on Algorithms, High Performance Computing, and Artificial Intelligence(AHPCAI 2023).Yinchuan：SPIE，2023，12941：1223-1231.
  11. [11] YANG F，MAO Q. Auto-Evaluation Model for the Prediction of Building Energy Consumption That Combines Modified Kalman Filtering and Long Short-Term Memory [J/OL].Sustainability，2023，15（22）：15749[2025-06-28].https://www.mdpi.com/2071-1050/15/22/15749.
  12. [12] TAO F H. Performance Analysis of Real Estate Management Entities Based on DEA Model and BP Neural Network Model [JO/L].Journal of Physics，2020，1744（2）：022025[2025-07-18].https://iopscience.iop.org/article/10.1088/1742-6596/1744/2/022025.
  13. [13] XU H J，LI S，FAN S D，et al. A new Inconsistent Context Fusion Algorithm Based on BP Neural Network and Modified DST [J].Mathematical Biosciences and Engineering，2021，18（2）：968-982.
  14. [14] ZHANG L，LIANG F Z，WAGNER N. Monitoring and Analysis of Athletes’ Local Body Movement Status Based on BP Neural Network [J].Journal of Intelligent & Fuzzy Systems，2021，40（2）：2325-2335.
  15. [15] KUMARI S，KUMAR N，RANA P S. Comparative Performance Study of Different Filtering Techniques with LSTM for the Prediction of Power Consumption in Smart Grid [J].IETE Journal of Research，2024，70（4）：3646-3663.
  16. [16] SADOWSKY J. Investigation of Signal Characteristics Using the Continuous Wavelet Transform [J].Johns Hopkins APL Technical Digest，1996，17（3）：258-269.
  17. [17] CHI D. Research on Electricity Consumption Forecasting Model Based on Wavelet Transform and Multi-Layer LSTM Model [J].Energy Reports，2022，8：220-228.
  18. [18] AHAJJAM M A，LICEA D B，ESSAYEH C，et al. MORED: A Moroccan Buildings’ Electricity Consumption Dataset [J/OL].Energies，2020，13（24）：6737[2025-05-18].https://www.mdpi.com/1996-1073/13/24/6737.
  19. [19] ERICK. Cantu, The Green City Solutions Group, GitHub [EB/OL].[2025-06-19].https://github.com/erickCantu/TheGreenCitySolutionsGroup.

About the authors

Tong Yuxing (1993-), male, Han ethnicity, from Wuhan, Hubei, engineer, master's degree, research interests include railway communication and information; Xiong Yi (1997-), male, Han ethnicity, from Wuhan, Hubei, assistant engineer, master's degree, research interests include communication engineering.