Ajay Kumar; Rainu Nandal; Kamaldeep Joshi

Multi-Horizon Deep Learning Framework for Sustainable Energy Forecasting in Smart Campuses: Enabling Energy Efficient Smart Educational Buildings

PDF (1869KB), PP.168-191

Views: 0 Downloads: 0

Author(s)

Ajay Kumar ¹ Rainu Nandal ^1,* Kamaldeep Joshi ¹

1. Department of Computer Science and Engineering, University Institute of Engineering and Technology, Maharshi Dayanand University, Rohtak, 124001, India

* Corresponding author.

DOI: https://doi.org/10.5815/ijmecs.2026.05.10

Received: 10 Oct. 2025 / Revised: 20 Dec. 2025 / Accepted: 13 Feb. 2026 / Published: 8 Oct. 2026

Index Terms

Smart Campuses, Energy Forecasting, Energy Efficiency, Deep Learning, Educational Infrastructure, Sustainable Energy Management.

Abstract

The accurate energy consumption forecast of educational institutions is critical for sustainable energy management and intelligent campus infrastructure. Classical forecasting models fail to simultaneously capture local temporal variations and long-range dependencies, thus diminishing their reliability across different time horizons. To overcome this, we propose an effective hybrid deep learning method that combines Convolutional Neural Networks (CNNs) with Long Short-Term Memory (LSTM) networks and domain-driven feature engineering to incorporate occupancy and seasonality patterns. The dataset from an office building (research support facility) was used to perform a series of short-term (05-minutes), medium-term (daily), and long-term (weekly) forecasts. Experimental results show that the proposed forecasting approach demonstrates greater precision than the ten baselines including machine learning and deep learning methods in short, medium, and long time horizons. In particular, the model with feature engineering improved Root Mean Squared Error (RMSE) by 4.71% and Mean Absolute Percentage Error (MAPE) by 3.99% in the short term and saw even greater improvements for medium-term forecasts, with RMSE improved by 6.08% and MAPE by 5.37%, and for long-term forecasts, with RMSE improved by 2.58% and MAPE by 0.33%, respectively. The outcomes here offer significant benefits to the science community, supporting the development of such smart campuses with a specific, understandable methodology for energy forecasting. The methodology enables the application of data-driven energy management, cost optimization, and sustainability planning in educational spaces. The findings also provide valuable insights for deploying Artificial Intelligence (AI)-driven infrastructure management systems in academia and institutional energy governance.

Cite This Paper

Ajay Kumar, Rainu Nandal, Kamaldeep Joshi, "Multi-Horizon Deep Learning Framework for Sustainable Energy Forecasting in Smart Campuses: Enabling Energy Efficient Smart Educational Buildings", International Journal of Modern Education and Computer Science(IJMECS), Vol.18, No.5, pp. 168-191, 2026. DOI:10.5815/ijmecs.2026.05.10

Reference

[1]M. Q. Raza, and A. Khosravi, “A review on artificial intelligence-based load demand forecasting techniques for smart grid and buildings,” Renew. Sustain. Energy Rev., vol. 50, pp. 1352–1372, 2015. doi:10.1016/j.rser.2015.04.065.
[2]N. Wei, C. Li, X. Peng, F. Zeng, X. Lu, “Conventional models and artificial intelligence-based models for energy consumption forecasting: a review,” J. Petroleum Sci. Eng., vol. 181, pp. 106187, 2019. doi:10.1016/j.petrol.2019.106187.
[3]D. Kaur, S. N. Islam, M. Mahmud, Z. Dong, “Energy forecasting in smart grid systems: a review of the state-of-the-art techniques,” arXiv Prepr, 2020. arXiv:2011.12598.
[4]L. Pérez-Lombard, J. Ortiz, and C. Pout, “A review on buildings energy consumption information,” Energy Build., vol. 40, no. 3, pp. 394–398, 2008. doi:10.1016/j.enbuild.2007.03.007.
[5]Y. Ji, K.J. Lomas, and M.J. Cook, “Hybrid ventilation for low energy building design in south China,” Build. Environ., vol. 44, pp. 2245–2255, 2009. doi:10.1016/j.buildenv.2009.02.015.
[6]S. Manu, Y. Shukla, R. Rawal, L.E. Thomas, and R. De Dear, “Field studies of thermal comfort across multiple climate zones for the subcontinent India Model for Adaptive Comfort (IMAC),” Build. Environ., vol. 98, pp. 55–70, 2016. doi:10.1016/j.buildenv.2015.12.019.
[7]E. Delzendeh, S. Wu, A. Lee, and Y. Zhou, “The impact of occupants’ behaviours on building energy analysis: A research review,” Renew. Sustain. Energy Rev., vol. 80, pp. 1061–1071, 2017. doi:10.1016/j.rser.2017.05.264.
[8]M. I. Ahmad, “Seasonal decomposition of electricity consumption data,” Rev. Integr. Bus. Econ. Res., vol. 6, no. 4, pp. 271, 2017
[9]H. Kaur, and S. Ahuja, “Time series analysis and prediction of electricity consumption of health care institution using ARIMA model,” In Proc. Sixth Int. Conf. Soft Comput. Probl. Solving, Singapore, pp. 347–358, 2017, December 2016. doi:10.1007/978-981-10-3325-4_35.
[10]K. Amarasinghe, D. L. Marino, and M. Manic, “Deep neural networks for energy load forecasting,” In 2017 IEEE 26th Int. Symp. Ind. Electron. (ISIE), Edinburgh, United Kingdom, pp. 1483–1488, June 2017. doi:10.1109/ISIE.2017.8001465.
[11]S. Nazir, A. Ab Aziz, J. Hosen, N. A. Aziz, G. R. Murthy, “Forecast energy consumption time-series dataset using multistep LSTM models,” J. Phys. Conf. Ser., vol.1933, no. 1, pp. 012054, 2021. doi:10.1088/1742-6596/1933/1/012054.
[12]J.G. Jetcheva, M. Majidpour, W.P. Chen, “Neural network model ensembles for building-level electricity load forecasts,” Energy Build., vol. 84, pp. 214–223, 2014. doi:10.1016/j.enbuild.2014.08.004
[13]M.A. Khairalla, X. Ning, N.T. AL-Jallad, M.O. El-Faroug, “Short-Term Forecasting for Energy Consumption through Stacking Heterogeneous Ensemble Learning Model,” Energies, vol. 11, pp. 1605, 2018. doi:10.3390/en11061605.
[14]J. Geng, , M.L. Huang, M.W. Li, W.C. Hong, “Hybridization of seasonal chaotic cloud simulated annealing algorithm in a SVR-based load forecasting model,” Neurocomputing, vol. 151, pp. 1362–1373, 2015. doi:10.1016/j.neucom.2014.10.055.
[15]G.F. Fan, L.L. Peng, W.C. Hong, and F. Sun, “Electric load forecasting by the SVR model with differential empirical mode decomposition and auto regression,” Neurocomputing, vol. 173, pp. 958–970, 2016. doi:10.1016/j.neucom.2015.08.051.
[16]Z. Wang, and R.S. Srinivasan, “A review of artificial intelligence based building energy prediction with a focus on ensemble prediction models,” In 2015 Winter simul. conf. (WSC), pp. 3438–3448, 2016. doi:10.1109/WSC.2015.7408504.
[17]P.W. Tien, S. Wei, J. Darkwa, et al, “Machine learning and deep learning methods for enhancing building energy efficiency and indoor environmental quality - a review,” Energy AI, vol. 10, pp. 100198, 2022. doi:10.1016/j.egyai.2022.100198.
[18]J. Lago, G. Marcjasz, B. De Schutter et al, “Forecasting day-ahead electricity prices: A review of state-of-the-art algorithms, best practices and an open-access benchmark,” Appl. Energy, vol. 293, no. 116, pp. 983, 2021. doi:10.1016/j.apenergy.2021.116983.
[19]J.S. Chou, and N.T. Ngo, “Time series analytics using sliding window metaheuristic optimization-based machine learning system for identifying building energy consumption patterns,” Appl. Energy, vol. 177, pp. 751–770, 2016. doi:10.1016/j.apenergy.2016.05.074.
[20]Y. Kim, H.G. Son, and S. Kim, “Short term electricity load forecasting for institutional buildings,” Energy Rep., vol. 5, pp. 1270–1280, 2019. doi:10.1016/j.egyr.2019.08.086.
[21]T. Pinto, I. Praça, Z. Vale, and J. Silva, “Ensemble learning for electricity consumption forecasting in office buildings,” Neurocomputing, vol. 423, pp. 747–755, 2021. doi:10.1016/j.neucom.2020.02.124.
[22]G. Chitalia, M. Pipattanasomporn, V. Garg, and S. Rahman, “Robust short-term electrical load forecasting framework for commercial buildings using deep recurrent neural networks,” Appl. Energy, vol. 278, pp. 115410, 2020. doi:10.1016/j.apenergy.2020.115410.
[23]F. Pallonetto, C. Jin, and E. Mangina, “Forecast electricity demand in commercial building with machine learning models to enable demand response programs,” Energy AI, vol. 7, pp. 100121, 2022. doi:10.1016/j.egyai.2021.100121.
[24]J. Hwang, D. Suh, and M.O. Otto, “Forecasting electricity consumption in commercial buildings using a machine learning approach,” Energies, 13, pp. 5885, 2020. doi:10.3390/en13225885.
[25]X. Wen, J. Liao, Q. Niu, N. Shen, and Y. Bao, “Deep learning-driven hybrid model for short-term load forecasting and smart grid information management”, Sci. Rep., vol. 14, pp. 13720, 2024. doi:10.1038/s41598-024-63262-x.
[26]H. Li, S. Li, Y. Wu, Y. Xiao, Z. Pan, and M. Liu, “Short-term power load forecasting for integrated energy system based on a residual and attentive LSTM-TCN hybrid network”. Front. Energy Res., vol. 12, pp. 1384142, 2024. doi:10.3389/fenrg.2024.1384142.
[27]J. Bai, W. Zhu, S. Liu, C. Ye, P. Zheng, and X. Wang, “A Temporal Convolutional Network–Bidirectional Long Short-Term Memory (TCN-BiLSTM) Prediction Model for Temporal Faults in Industrial Equipment”, Appl. Sci., vol.15, no. 4, p. 1702, 2025. https://doi.org/10.3390/app15041702
[28]B. Parizad, H. Ranjbarzade, A. Jamali, and H. Khayyam, “An Intelligent Hybrid Machine Learning Model for Sustainable Forecasting of Home Energy Demand and Electricity Price”, Sustainability, vol. 16, no. 6, pp. 2328, 2024. doi:10.3390/su16062328.
[29]Z. Bai, “Residential electricity prediction based on GA-LSTM modeling”, Energy Rep., vol. 11, pp. 6223-6232, 2024. doi:10.1016/j.egyr.2024.06.010.
[30]Y. Özüpak, and S. Mansurov, “Optimizing electricity demand forecasting with a novel RNN-LSTM hybrid model”, Energy Sources B: Econ. Plan. Policy, vol. 20, no. 1, 2025. doi:10.1080/15567249.2025.2531448.
[31]D. Mariano-Hernández, et al., “Comparative study of continuous hourly energy consumption forecasting strategies with small data sets to support demand management decisions in buildings,” Energy Sci. Eng., vol. 10, pp. 4694–4707, 2022. doi:10.1002/ese3.1298.
[32]R. Spencer, S. Ranathunga, M. Boulic, A.H. van Heerden, and T. Susnjak, “Transfer learning on transformers for building energy consumption forecasting—A comparative study”. Energy Build., vol. 336, pp. 115632, 2025. doi:10.1016/j.enbuild.2025.115632.
[33]D. Fernández-Martínez, and M.A. Jaramillo-Morán, “Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building with Recurrent Neural Networks and Empirical Mode Decomposition,” Sensors, vol. 22, pp. 3664, 2022. doi:10.3390/s22103664.
[34]E. Skomski, J.Y. Lee, W. Kim, Chandan, S. Katipamula, and B. Hutchinson, “Sequence-to-sequence neural networks for short-term electrical load forecasting in commercial office buildings,” Energy Build., vol. 226, pp. 110350, 2020. doi:10.1016/j.enbuild.2020.110350.
[35]I. Karijadi, and S.Y. Chou, “A hybrid RF-LSTM based on CEEMDAN for improving the accuracy of building energy consumption prediction,” Energy Build., vol. 259, pp. 111908, 2022. doi:10.1016/j.enbuild.2022.111908.
[36]L. Lei, S.Shao, L. Liang, “An evolutionary deep learning model based on EWKM, random forest algorithm, SSA, and BiLSTM for building energy consumption prediction,” Energy, vol. 288, pp. 129795, 2024. doi:10.1016/j.energy.2023.129795.
[37]Z. Ding, W. Chen, T. Hu, X. Xu, “Evolutionary double attention-based long short-term memory model for building energy prediction: Case study of a green building,” Appl. Energy., vol. 288, pp. 116660, 2021. doi:10.1016/j.apenergy.2021.116660.
[38]B. Doherty, and K. Trenbath, “Data for: Device-level plug load disaggregation in a zero energy office building and opportunities for energy savings,” Mendeley Data., vol. 06, 2019, doi: 10.17632/g392vt7db9.1.
[39]J. Sessa, and D. Syed, “Techniques to deal with missing data,” In Proc. 5th Int. Conf. Electron. Devices, Syst. Appl. (ICEDSA), Sarawak, Malaysia. pp. 1–4, December 2016. doi: 10.1109/ICEDSA.2016.7818486.
[40]S. V. Khond, “Effect of data normalization on accuracy and error of fault classification for an electrical distribution system,” Smart Sci., vol. 8, no. 3, pp. 117–124, 2020. doi: 10.1080/23080477.2020.1799135.
[41]S. Bai, V. Koltun, and J.Z. Kolter, “Multiscale deep equilibrium models.” NIPS'20: Proc. 34th Int. Conf. Neural Inf. Process. Syst.., Red Hook, United States, vol. 33, pp. 5238-5250, December 2020.
[42]S. Kasmaiee, S. Kasmaiee, and M. Homayounpour, “Correcting spelling mistakes in Persian texts with rules and deep learning methods,” Sci. rep., vol. 13, no. 1, pp. 19945, 2023. doi:10.1038/s41598-023-47295-2.
[43]N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, "Dropout: A simple way to prevent neural networks from overfitting," Journal of Machine Learning Research, vol. 15, pp. 1929–1958, 2014.
[44]B. G. Marcot, and Hanea, A. M., “What is an optimal value of k in k-fold cross-validation in discrete Bayesian network analysis?,” Comput. Statist., vol. 36, no. 3, pp. 2009–2031, 2021. doi: 10.1007/s00180-020-00999-9.
[45]B. Devanathan, P. Manelllore, S.A. Lakshmanan, S. Kayalvizhi,, and A. Suyampulingam, “AI-Based Energy Consumption Forecasting in Smart Homes Using Multi-Algorithm Analysis Across Temporal Resolutions, Results Eng., vol. 29, pp. 109272. 2026. doi:10.1016/j.rineng.2026.109272.
[46]B.K. Hammerschmitt, M.V.H. Rambo, A.D.S. Leone, L.M. Iantorno, H.B. Schiavon, D.P. Corrêa, P. Lissa, M. Keane, and R.J. Riella, “Deep Learning for Residential Electrical Energy Consumption Forecasting: A Hybrid Framework with Multiscale Temporal Analysis and Weather Integration”, Energies, vol. 18, no. 22, pp. 5885, 2025. doi:10.3390/en18225885.
[47]D. Syed, S. S. Refaat, H. Abu-Rub, “Performance evaluation of distributed machine learning for load forecasting in smart grids,” In Proc. Cybern. Informat. (K&I)., pp. 1–6, 2020. doi: 10.1109/KI48306.2020.9039797.

International Journal of Modern Education and Computer Science (IJMECS)