TAccurate short-term electricity consumption forecasting is essential for operational planning, reserve allocation, and energy management in modern power systems. This study investigates the performance of recurrent and hybrid attention-based deep learning architectures for short-term electricity consumption forecasting, a sase study in Togo. The proposed framework integrates electricity consumption, meteorological, demographic, and temporal information in order to capture both intrinsic temporal dependencies and exogenous influences affecting electricity demand. Four forecasting architectures were evaluated using a weekly temporal window of 168 hours: LSTM-only, LSTM-decoder, LSTM-attention, and a hybrid LSTM--Multi-Head Attention--LSTM model. The experiments included multi-seed evaluation, ablation study, robustness analysis, and statistical comparison using the Wilcoxon signed-rank test. The results show that all models achieved extremely high forecasting accuracy, with coefficients of determination exceeding 0.9998 and MAPE values below 0.004%. The hybrid architectures slightly improved average forecasting performance, while the standalone LSTM model remained highly competitive. The robustness analysis revealed strong sensitivity to noisy inputs but moderate degradation under missing-data conditions. Statistical analysis indicated that the performance differences between architectures were not statistically significant at the 5% level. Overall, the study demonstrates that recurrent deep learning architectures provide reliable and operationally relevant solutions for electricity consumption forecasting in highly structured energy demand series.
| Published in | Science Journal of Energy Engineering (Volume 14, Issue 3) |
| DOI | 10.11648/j.sjee.20261403.11 |
| Page(s) | 65-82 |
| Creative Commons |
This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited. |
| Copyright |
Copyright © The Author(s), 2026. Published by Science Publishing Group |
Model, LSTM, Multi-Head Attention, Energy, Prediction, Consumption
| [1] | Dinata, S., Azka, M., Muliady, F., et al., Short-Term Load Forecasting Double Seasonal ARIMA Methods: An Evaluation Based on Mahakam-East Kalimantan Data, AIP Conference Proceedings, vol. 2268, article 020004, 2020, |
| [2] |
Open-Meteo, Archive Weather API, Open-Meteo Documentation, 2024, Available:
https://archive-api.open-meteo.com/ Accessed 16 May 2026 |
| [3] |
World Bank, World Development Indicators, World Bank Open Data, 2024, Available:
https://donnees.banquemondiale.org/ Accessed 16 May 2026 |
| [4] | Nguyen, Q., Nguyen, N., Tran Ngoc, T., et al., Online SARIMA Applied for Short-Term Electricity Load Forecasting, Research Square, pp. 1003-1019, 2023, |
| [5] | Ajlouni, S. A., Using ARIMA Model to Forecast Electricity Load in Jordan, Jordan Journal of Earth and Environmental Sciences, vol. 11, no. 2, pp. 127-139, 2024, |
| [6] | Matos, M., Almeida, J., Gonçalves, P., et al., A Machine Learning-Based Electricity Consumption Forecast and Management System for Renewable Energy Communities, Energies, vol. 17, article 630, 2024, |
| [7] | Fan, S. and Hyndman, R. J., Short-Term Load Forecasting Based on a Semi-Parametric Additive Model, IEEE Transactions on Power Systems, vol. 27, pp. 134-141, 2012, |
| [8] | Hong, T., Pinson, P., and Fan, S., Global Energy Forecasting Competition 2012, International Journal of Forecasting, vol. 30, pp. 357-363, 2014, |
| [9] | Various Authors, Long Short-Term Memory Networks: A Comprehensive Survey, Engineering Proceedings, vol. 6, article 215, 2025, |
| [10] | Hong, T. and Fan, S., Probabilistic Energy Forecasting, International Journal of Forecasting, vol. 32, pp. 896-913, 2016, |
| [11] | Weron, R., Electricity Price Forecasting: A Review, International Journal of Forecasting, vol. 30, pp. 1030-1081, 2014, |
| [12] | Box, G. E. P., Jenkins, G. M., Reinsel, G. C., and Ljung, G. M., Time Series Analysis: Forecasting and Control, John Wiley & Sons, 5th ed., 712 pages, 2015 |
| [13] | Hochreiter, S. and Schmidhuber, J., Long Short-Term Memory, Neural Computation, vol. 9, pp. 1735-1780, 1997, |
| [14] | Bokovi, Y., Moyème Kabe, S., Sedzro Kwami, S., Takouda, P., and Lare, Y., Machine Learning Electrical Load Forecasting: An Application in Microgrid Energy Consumption with Adaboost Regressor Approach and a Comparative Study with Hybrid Method Based on LSTM and MLP Approaches, Journal of Sustainable Development of Energy, Water and Environment Systems, vol. 13, article 1130606, 2025, |
| [15] | Suganthi, L. and Samuel, A. A., Energy Models for Demand Forecasting-A Review, Renewable and Sustainable Energy Reviews, vol. 16, pp. 1223-1240, 2012, |
| [16] | Kong, W., Dong, Z. Y., Jia, Y., et al., Short-Term Residential Load Forecasting Based on LSTM, IEEE Transactions on Smart Grid, vol. 10, pp. 841-851, 2019, |
| [17] | Vaswani, A., Shazeer, N., Parmar, N., et al., Attention Is All You Need, Advances in Neural Information Processing Systems (NeurIPS), 2017, |
| [18] | Ozcan, A., Catal, C., and Kasif, A., Energy Load Forecasting Using Attention RNN, Sensors, vol. 21, article 7115, 2021, |
| [19] | Ding, S., He, D., and Liu, G., Improving Short-Term Load Forecasting with CNN and Multi-Head Attention, Electronics, vol. 13, article 5023, 2024, |
| [20] | Hua, Q., Fan, Z., Mu, W., et al., CNN-GRU with Attention Mechanism, Energies, vol. 18, article 106, 2025, |
| [21] | Wu, H., Xu, J., Wang, J., and Long, M., Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting, arXiv preprint, 2021, Available: |
| [22] | Semekonawo, K. P. and Kam, S., Electricity Demand Forecasting in West Africa, Journal of Energy Research and Reviews, vol. 12, pp. 26-36, 2022, |
| [23] | Palanga, E. T. G., Bara, K. K. A., and Barate, M., Modeling of Electric Energy Consumption in Humid Zones Using Selected Meteorological Variables Through XGBoost, ANFIS, and RNN Approaches, American Journal of Energy Engineering, vol. 13, pp. 189-212, 2025, |
| [24] | Liu, J., Short-Term Load Forecasting of Electric Power System Based on Meteorological Factors, Proceedings of ICMMBE 2016, 2016, |
| [25] | Cheng, Y., et al., Short-Term Load Forecasting Considering Improved Cumulative Effect of Hourly Temperature, Electric Power Systems Research, 2021, |
| [26] | Andoh, P. Y., Sekyere, C. K., et al., Forecasting Electricity Demand in Ghana, Journal of Applied Engineering and Technological Science, vol. 3, 2021, |
| [27] | Adjamagbo, C., Salami, A., Bokovi, Y., et al., Forecasting of Electric Energy Consumption in Togo, International Journal of Advanced Research, vol. 8, pp. 22-28, 2020, |
| [28] | Amega, K., Moumouni, Y., and Lare, Y., Electricity Consumption in Lom'e, International Journal of Energy and Power Engineering, vol. 10, pp. 141-150, 2021, |
APA Style
Apeke, S., Bokovi, Y., Gbafa, K. S., Kolegain, K., Ouro-Djobo, S. (2026). Comparative Analysis of LSTM and Hybrid Attention-Based Architectures for Short-Term Electricity Consumption Forecasting: A Case Study in Togo. Science Journal of Energy Engineering, 14(3), 65-82. https://doi.org/10.11648/j.sjee.20261403.11
ACS Style
Apeke, S.; Bokovi, Y.; Gbafa, K. S.; Kolegain, K.; Ouro-Djobo, S. Comparative Analysis of LSTM and Hybrid Attention-Based Architectures for Short-Term Electricity Consumption Forecasting: A Case Study in Togo. Sci. J. Energy Eng. 2026, 14(3), 65-82. doi: 10.11648/j.sjee.20261403.11
Share This Article
Twitter
Linked In
Facebook
Copy
Download@article{10.11648/j.sjee.20261403.11,
author = {Sena Apeke and Yao Bokovi and Kodjovi Senanou Gbafa and Komlan Kolegain and Sanoussi Ouro-Djobo},
title = {Comparative Analysis of LSTM and Hybrid Attention-Based Architectures for Short-Term Electricity Consumption Forecasting: A Case Study in Togo},
journal = {Science Journal of Energy Engineering},
volume = {14},
number = {3},
pages = {65-82},
doi = {10.11648/j.sjee.20261403.11},
url = {https://doi.org/10.11648/j.sjee.20261403.11},
eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.sjee.20261403.11},
abstract = {TAccurate short-term electricity consumption forecasting is essential for operational planning, reserve allocation, and energy management in modern power systems. This study investigates the performance of recurrent and hybrid attention-based deep learning architectures for short-term electricity consumption forecasting, a sase study in Togo. The proposed framework integrates electricity consumption, meteorological, demographic, and temporal information in order to capture both intrinsic temporal dependencies and exogenous influences affecting electricity demand. Four forecasting architectures were evaluated using a weekly temporal window of 168 hours: LSTM-only, LSTM-decoder, LSTM-attention, and a hybrid LSTM--Multi-Head Attention--LSTM model. The experiments included multi-seed evaluation, ablation study, robustness analysis, and statistical comparison using the Wilcoxon signed-rank test. The results show that all models achieved extremely high forecasting accuracy, with coefficients of determination exceeding 0.9998 and MAPE values below 0.004%. The hybrid architectures slightly improved average forecasting performance, while the standalone LSTM model remained highly competitive. The robustness analysis revealed strong sensitivity to noisy inputs but moderate degradation under missing-data conditions. Statistical analysis indicated that the performance differences between architectures were not statistically significant at the 5% level. Overall, the study demonstrates that recurrent deep learning architectures provide reliable and operationally relevant solutions for electricity consumption forecasting in highly structured energy demand series.},
year = {2026}
}
TY - JOUR T1 - Comparative Analysis of LSTM and Hybrid Attention-Based Architectures for Short-Term Electricity Consumption Forecasting: A Case Study in Togo AU - Sena Apeke AU - Yao Bokovi AU - Kodjovi Senanou Gbafa AU - Komlan Kolegain AU - Sanoussi Ouro-Djobo Y1 - 2026/08/21 PY - 2026 N1 - https://doi.org/10.11648/j.sjee.20261403.11 DO - 10.11648/j.sjee.20261403.11 T2 - Science Journal of Energy Engineering JF - Science Journal of Energy Engineering JO - Science Journal of Energy Engineering SP - 65 EP - 82 PB - Science Publishing Group SN - 2376-8126 UR - https://doi.org/10.11648/j.sjee.20261403.11 AB - TAccurate short-term electricity consumption forecasting is essential for operational planning, reserve allocation, and energy management in modern power systems. This study investigates the performance of recurrent and hybrid attention-based deep learning architectures for short-term electricity consumption forecasting, a sase study in Togo. The proposed framework integrates electricity consumption, meteorological, demographic, and temporal information in order to capture both intrinsic temporal dependencies and exogenous influences affecting electricity demand. Four forecasting architectures were evaluated using a weekly temporal window of 168 hours: LSTM-only, LSTM-decoder, LSTM-attention, and a hybrid LSTM--Multi-Head Attention--LSTM model. The experiments included multi-seed evaluation, ablation study, robustness analysis, and statistical comparison using the Wilcoxon signed-rank test. The results show that all models achieved extremely high forecasting accuracy, with coefficients of determination exceeding 0.9998 and MAPE values below 0.004%. The hybrid architectures slightly improved average forecasting performance, while the standalone LSTM model remained highly competitive. The robustness analysis revealed strong sensitivity to noisy inputs but moderate degradation under missing-data conditions. Statistical analysis indicated that the performance differences between architectures were not statistically significant at the 5% level. Overall, the study demonstrates that recurrent deep learning architectures provide reliable and operationally relevant solutions for electricity consumption forecasting in highly structured energy demand series. VL - 14 IS - 3 ER -
Engineering Science Research Laboratory, National Polytechnic School, Lome, Togo; Regional Center of Excellence for Electricity Control, Solar Energy Laboratory, University of Lome, Lome, Togo
Engineering Science Research Laboratory, National Polytechnic School, Lome, Togo; Regional Center of Excellence for Electricity Control, Solar Energy Laboratory, University of Lome, Lome, Togo
Laboratory of Structures and Materials Mechanics, National Polytechnic School, Lome, Togo
Regional Center of Excellence for Electricity Control, Solar Energy Laboratory, University of Lome, Lome, Togo
Regional Center of Excellence for Electricity Control, Solar Energy Laboratory, University of Lome, Lome, Togo; Solar Energy Laboratory, Department of Physics, Universite of Lome, Lome, Togo
Information