Accurate forecasting of electrical, heating, and cooling demands is crucial for the optimal design, operation, and coordinated planning of multi-energy coupling systems. In this study, we present a short-term multi-energy load forecasting method for integrated energy systems based on a temporal convolutional network and multi-task learning. First, the Pearson correlation coefficient was used to analyze the correlations among meteorological variables. Then, the maximum overlap discrete wavelet transform (MODWT) module decomposed the input time-series data into components at different frequency levels. These decomposed time series datasets were input into the multi-task learning model for sparse sharing. Subsequently, the learned time series datasets of different frequencies were input into the temporal convolutional network for temporal scale feature extraction. To improve prediction accuracy, the improved energy valley optimizer (IEVO) was introduced to optimize the model parameters. The feature vectors output by the optimized temporal convolutional network were then fed into the corresponding Transformer models for adaptive weight adjustment. Finally, the MODWT module reconstructed the predictions at different frequency levels to obtain the final multi-energy load forecasting results.
Citation: He Jiang, Tianhui Shi, Lixin Yuan. Short-term multi-energy load forecasting of integrated energy systems based on temporal convolutional network and multi-task learning[J]. AIMS Energy, 2026, 14(4): 874-904. doi: 10.3934/energy.2026035
Accurate forecasting of electrical, heating, and cooling demands is crucial for the optimal design, operation, and coordinated planning of multi-energy coupling systems. In this study, we present a short-term multi-energy load forecasting method for integrated energy systems based on a temporal convolutional network and multi-task learning. First, the Pearson correlation coefficient was used to analyze the correlations among meteorological variables. Then, the maximum overlap discrete wavelet transform (MODWT) module decomposed the input time-series data into components at different frequency levels. These decomposed time series datasets were input into the multi-task learning model for sparse sharing. Subsequently, the learned time series datasets of different frequencies were input into the temporal convolutional network for temporal scale feature extraction. To improve prediction accuracy, the improved energy valley optimizer (IEVO) was introduced to optimize the model parameters. The feature vectors output by the optimized temporal convolutional network were then fed into the corresponding Transformer models for adaptive weight adjustment. Finally, the MODWT module reconstructed the predictions at different frequency levels to obtain the final multi-energy load forecasting results.
| [1] |
Guo X, Xiao B, Song L (2020) Emission reduction and energy-intensity enhancement: The expected and unexpected consequences of China's coal consumption constraint policy. J Clean Prod 271: 122691. https://doi.org/10.1016/j.jclepro.2020.122691 doi: 10.1016/j.jclepro.2020.122691
|
| [2] |
Li J, Liu J, Yan P, et al. (2021) Operation optimization of integrated energy system under a renewable energy dominated future scene considering both independence and benefit: A review. Energies 14: 1103https://doi.org/10.3390/en14041103 doi: 10.3390/en14041103
|
| [3] |
Guo Z, Zhang R, Wang L, et al. (2021) Optimal operation of regional integrated energy system considering demand response. Appl Therm Eng 191: 116860. https://doi.org/10.1016/j.applthermaleng.2021.116860 doi: 10.1016/j.applthermaleng.2021.116860
|
| [4] |
Fan H, Wang C, Liu L, et al. (2022) Review of uncertainty modeling for optimal operation of integrated energy system. Front Energy Res 9: 641337. https://doi.org/10.3389/fenrg.2021.641337 doi: 10.3389/fenrg.2021.641337
|
| [5] |
Liu X, Liu W, Zhou W, et al. (2024) Multi-energy load forecasting for IES considering meteorological causation and repetition cycles. Electr Pow Syst Res 237: 110977. https://doi.org/10.1016/j.epsr.2024.110977 doi: 10.1016/j.epsr.2024.110977
|
| [6] |
Tan M, Liao C, Chen J, et al. (2023) A multi-task learning method for multi-energy load forecasting based on synthesis correlation analysis and load participation factor. Appl Energy 343: 121177. https://doi.org/10.1016/j.apenergy.2023.121177 doi: 10.1016/j.apenergy.2023.121177
|
| [7] |
Wang J, Lu S, Wang S, et al. (2021) A review on extreme learning machine. Multimed Tools Appl 81: 41611–41660. https://doi.org/10.1007/s11042-021-11007-7 doi: 10.1007/s11042-021-11007-7
|
| [8] |
Noori N, Kalin L, Isik S (2020) Water quality prediction using SWAT-ANN coupled approach. J Hydrol 590: 125220. https://doi.org/10.1016/j.jhydrol.2020.125220 doi: 10.1016/j.jhydrol.2020.125220
|
| [9] | Pasupuleti VR, Uhasri, Kalyan P, et al. (2020) Air quality prediction of data log by machine learning. 2020 6th International Conference on Advanced Computing and Communication Systems (ICACCS), Coimbatore, India, 1395–1399. https://doi.org/10.1109/icaccs48705.2020.9074431 |
| [10] |
Yazici I, Beyca OF, Delen D (2022) Deep-learning-based short-term electricity load forecasting: A real case application. Eng Appl Artif Intel 109: 104645. https://doi.org/10.1016/j.engappai.2021.104645 doi: 10.1016/j.engappai.2021.104645
|
| [11] |
Wang S, Zhang Z (2021) Short-Term multiple load forecasting model of regional integrated energy system based on QWGRU-MTL. Energies 14: 6555. https://doi.org/10.3390/en14206555 doi: 10.3390/en14206555
|
| [12] |
Cen S, Lim CG (2024) Multi-Task learning of the PatchTCN-TST model for short-term multi-load energy forecasting considering indoor environments in a smart building. IEEE Access 12: 19553–19568. https://doi.org/10.1109/access.2024.3355448 doi: 10.1109/access.2024.3355448
|
| [13] |
Qiu Y, Wang S, Zhang S, et al. (2023) A novel gated dual convolutional neural network model with autoregressive method and attention mechanism for probabilistic load forecasting. Appl Intell 53: 20256–20271. https://doi.org/10.1007/s10489-023-04589-2 doi: 10.1007/s10489-023-04589-2
|
| [14] |
Zhang Z, Zhang C, Dong Y, et al. (2025) Bi-directional gated recurrent unit enhanced twin support vector regression with seasonal mechanism for electric load forecasting. Knowledge-Based Syst 310: 112943. https://doi.org/10.1016/j.knosys.2024.112943 doi: 10.1016/j.knosys.2024.112943
|
| [15] |
Wang Q, Xu F, He J, et al. (2026) A new fusion model for enhanced ultra-short-term offshore wind power forecasting. Renewable Energy 256: 123876. https://doi.org/10.1016/j.renene.2025.123876 doi: 10.1016/j.renene.2025.123876
|
| [16] |
Xu F, Wang Q, Du X, et al. (2026) A novel hybrid short-term and ultra-short-term wind power forecasting method based on Weather Research and Forecasting: WRF-iTransformer-PSO. Energy 353: 140955. https://doi.org/10.1016/j.energy.2026.140955 doi: 10.1016/j.energy.2026.140955
|
| [17] |
He J, Liu QY, Hu YL, et al. (2026) Short-term regional EV charging load forecasting based on GAT and GRU with trip distribution estimation. Appl Energy 406: 127262. https://doi.org/10.1016/j.apenergy.2025.127262 doi: 10.1016/j.apenergy.2025.127262
|
| [18] |
Shang Y, Shang WL, Cui D, et al. (2026) Spatio-temporal data fusion framework based on large language model for enhanced prediction of electric vehicle charging demand in smart grid management. Inf Fusion 126: 103692. https://doi.org/10.1016/j.inffus.2025.103692 doi: 10.1016/j.inffus.2025.103692
|
| [19] |
Liu M, Qin H, Cao R, et al. (2022) Short-Term load forecasting based on improved TCN and DenseNet. IEEE Access 10: 115945–115957. https://doi.org/10.1109/access.2022.3218374 doi: 10.1109/access.2022.3218374
|
| [20] | Hong Q, Meng F, Maldonado F (2024) Advancing long-term multi-energy load forecasting with patchformer: A patch and transformer-based approach. arXiv. https://doi.org/10.48550/arXiv.2404.10458 |
| [21] |
Chen B, Wang Y (2021) Short-Term electric load forecasting of integrated energy system considering nonlinear synergy between different loads. IEEE Access 9: 43562–43573. https://doi.org/10.1109/access.2021.3066915 doi: 10.1109/access.2021.3066915
|
| [22] |
Imani M, Ghassemian H (2019) Residential load forecasting using wavelet and collaborative representation transforms. Appl Energy 253: 113505. https://doi.org/10.1016/j.apenergy.2019.113505 doi: 10.1016/j.apenergy.2019.113505
|
| [23] |
Fan GF, Han YY, Li JW, et al. (2024) A hybrid model for deep learning short-term power load forecasting based on feature extraction statistics techniques. Expert Syst Appl 238: 122012. https://doi.org/10.1016/j.eswa.2023.122012 doi: 10.1016/j.eswa.2023.122012
|
| [24] | Zhang T, Zhang J, Xue H, et al. (2021) Prediction of distribution network operation trend based on the secondary modal decomposition and LSTM-MFO algorithm. 2021 6th Asia Conference on Power and Electrical Engineering (ACPEE), Chongqing, China, 520–525. https://doi.org/10.1109/acpee51499.2021.9436870 |
| [25] |
Huang Q, Li J, Zhu M (2020) An improved convolutional neural network with load range discretization for probabilistic load forecasting. Energy 203: 117902. https://doi.org/10.1016/j.energy.2020.117902 doi: 10.1016/j.energy.2020.117902
|
| [26] |
Spoladore A, Borelli D, Devia F, et al. (2016) Model for forecasting residential heat demand based on natural gas consumption and energy performance indicators. Appl Energy 182: 488–499. https://doi.org/10.1016/j.apenergy.2016.08.122 doi: 10.1016/j.apenergy.2016.08.122
|
| [27] |
Ding Y, Zhang Q, Yuan T (2017) Research on short-term and ultra-short-term cooling load prediction models for office buildings. Energy Build 154: 254–267. https://doi.org/10.1016/j.enbuild.2017.08.077 doi: 10.1016/j.enbuild.2017.08.077
|
| [28] |
Li Y, Tian X, Liu T, et al. (2018) On better exploring and exploiting task relationships in multitask learning: Joint model and feature learning. IEEE Trans Neural Netw Learn Syst 29: 1975–1985. https://doi.org/10.1109/tnnls.2017.2690683 doi: 10.1109/tnnls.2017.2690683
|
| [29] |
Hou J, Meng Y, Li Z, et al. (2024) Multi-energy load forecasting in integrated energy systems based on comprehensive correlation index and SA-BiGRU network. Electr Power Constr 45: 118–130. https://doi.org/10.12204/j.issn.1000-7229.2024.05.012 doi: 10.12204/j.issn.1000-7229.2024.05.012
|
| [30] |
Tan Z, De G, Li M, et al. (2020) Combined electricity-heat-cooling-gas load forecasting model for integrated energy system based on multi-task learning and least square support vector machine. J Clean Prod 248: 119252. https://doi.org/10.1016/j.jclepro.2019.119252 doi: 10.1016/j.jclepro.2019.119252
|
| [31] |
You W, Guo D, Wu Y, et al. (2023) Multiple load forecasting of integrated energy system based on sequential-parallel hybrid ensemble learning. Energies 16: 3268. https://doi.org/10.3390/en16073268 doi: 10.3390/en16073268
|
| [32] | Yu X, Peng W, Xue D, et al. (2021) An improved spatial-temporal network based on residual correction and evolutionary algorithm for water quality prediction. In: Tan Y, Shi Y (eds) Advances in Swarm Intelligence. ICSI 2021. Lecture Notes in Computer Science, Springer, Cham, 12690. https://doi.org/10.1007/978-3-030-78811-7_46 |
| [33] |
Azad MA, Sajid I, Lu S, et al. (2023) Energy Valley Optimizer (EVO) for tracking the global maximum power point in a solar PV system under shading. Processes 11: 2986. https://doi.org/10.3390/pr11102986 doi: 10.3390/pr11102986
|
| [34] |
Fathy A (2023) Efficient energy valley optimization approach for reconfiguring thermoelectric generator system under non-uniform heat distribution. Renewable Energy 217: 119177. https://doi.org/10.1016/j.renene.2023.119177 doi: 10.1016/j.renene.2023.119177
|
| [35] |
Wang J, Han L, Zhang X, et al. (2023) Electrical load forecasting based on variable T-distribution and dual attention mechanism. Energy 283: 128569. https://doi.org/10.1016/j.energy.2023.128569 doi: 10.1016/j.energy.2023.128569
|
| [36] |
Azizi M, Aickelin U, Khorshidi HA, et al. (2023) Energy valley optimizer: A novel metaheuristic algorithm for global and engineering optimization. Sci Rep 13: 226. https://doi.org/10.1038/s41598-022-27344-y doi: 10.1038/s41598-022-27344-y
|
| [37] |
Jiang L, Wang X, Li W, et al. (2021) Hybrid multitask multi-information fusion deep learning for household short-term load forecasting. IEEE Trans Smart Grid 12: 5362–5372. https://doi.org/10.1109/tsg.2021.3091469 doi: 10.1109/tsg.2021.3091469
|
| [38] |
Li K, Duan P, Cao X, et al. (2024) A multi-energy load forecasting method based on complementary ensemble empirical model decomposition and composite evaluation factor reconstruction. Appl Energy 365: 123283. https://doi.org/10.1016/j.apenergy.2024.123283 doi: 10.1016/j.apenergy.2024.123283
|
| [39] | Gopali S, Abri F, Siami-Namini S, et al. (2021) A comparison of TCN and LSTM models in detecting anomalies in time series data. 2021 IEEE International Conference on Big Data (Big Data), Orlando, FL, USA, 2415–2420. https://doi.org/10.1109/BigData52589.2021.9671488 |
| [40] |
Gu Y, Jazizadeh F, Wang X (2025) Toward large energy models: A comparative study of Transformers' efficacy for energy forecasting. Appl Energy 384: 125358. https://doi.org/10.1016/j.apenergy.2025.125358 doi: 10.1016/j.apenergy.2025.125358
|
| [41] |
Xu S, Yang D (2025) WLENet: An workload-aware physician recommendation system with patient collective learning. J Supercomput 81: 1589. https://doi.org/10.1007/s11227-025-08091-z doi: 10.1007/s11227-025-08091-z
|
| [42] |
Geremicca F, Fascetti A, Brigham JC, et al. (2024) Urban metabolism and digital twin technologies for a sustainable built environment: Towards a framework for a campus application. IOP Conference Series: Earth and Environment 1363: 012081. https://doi.org/10.1088/1755-1315/1363/1/012081 doi: 10.1088/1755-1315/1363/1/012081
|
| [43] |
Gebeyhu B (2024) Power data access viewer-based meteorological drought analysis and rainfall variability in the Nile River Basin. Adv Meteorol, 2024: 9985773. https://doi.org/10.1155/2024/9985773 doi: 10.1155/2024/9985773
|
Figures(11) / Tables(4)
He Jiang, Tianhui Shi, Lixin Yuan. Short-term multi-energy load forecasting of integrated energy systems based on temporal convolutional network and multi-task learning[J]. AIMS Energy, 2026, 14(4): 874-904. doi: 10.3934/energy.2026035
DownLoad: