Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention
Accurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems. However, due to differences in response characteristics and load patterns among electrical, heating, and cooling loads, multi-energy load forecasting faces the challenges of heterogeneo...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Energy and AI |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S266654682500093X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1839626637924106240 |
|---|---|
| author | Yunfei Zhang Jun Shen Jian Li Mingzhe Yu Xu Chen Ziyong Yin |
| author_facet | Yunfei Zhang Jun Shen Jian Li Mingzhe Yu Xu Chen Ziyong Yin |
| author_sort | Yunfei Zhang |
| collection | DOAJ |
| description | Accurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems. However, due to differences in response characteristics and load patterns among electrical, heating, and cooling loads, multi-energy load forecasting faces the challenges of heterogeneous time scales and imbalanced forecasting accuracy across load types. To address these challenges, this paper proposes a multi-task learning model with stacked cross-attention. This model incorporates a time scale alignment module to align the time scales of different loads, and employs Informer encoders as experts to extract load-specific features. Stacked cross-attention as the soft sharing mechanism dynamically fuses expert features at the sequence level, enhancing inter-task collaboration and adaptability. This design improves the overall accuracy of multi-energy load forecasting with mixed time scales while reducing forecasting imbalance across load types. Comparison results demonstrate that the model with the stacked cross-attention achieves the best forecasting performance and lowers the imbalance index by 79.17 %. Furthermore, the experts based on Informer encoders yield over a 30.09 % MAPE reduction compared to alternative expert architectures. Compared to the multi-gate mixture-of-experts based models, classical forecasting models, and recent advanced models, the proposed model achieves superior forecasting accuracy, validating its effectiveness and advancement. |
| format | Article |
| id | doaj-art-9fcea888c87e4a0babd40d84a307a2bc |
| institution | Matheson Library |
| issn | 2666-5468 |
| language | English |
| publishDate | 2025-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Energy and AI |
| spelling | doaj-art-9fcea888c87e4a0babd40d84a307a2bc2025-07-17T04:45:03ZengElsevierEnergy and AI2666-54682025-09-0121100561Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attentionYunfei Zhang0Jun Shen1Jian Li2Mingzhe Yu3Xu Chen4Ziyong Yin5School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR China; State Key Laboratory of Power System Operation and Control, Tsinghua University, Beijing 100084, PR China; Corresponding author.School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, PR ChinaAccurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems. However, due to differences in response characteristics and load patterns among electrical, heating, and cooling loads, multi-energy load forecasting faces the challenges of heterogeneous time scales and imbalanced forecasting accuracy across load types. To address these challenges, this paper proposes a multi-task learning model with stacked cross-attention. This model incorporates a time scale alignment module to align the time scales of different loads, and employs Informer encoders as experts to extract load-specific features. Stacked cross-attention as the soft sharing mechanism dynamically fuses expert features at the sequence level, enhancing inter-task collaboration and adaptability. This design improves the overall accuracy of multi-energy load forecasting with mixed time scales while reducing forecasting imbalance across load types. Comparison results demonstrate that the model with the stacked cross-attention achieves the best forecasting performance and lowers the imbalance index by 79.17 %. Furthermore, the experts based on Informer encoders yield over a 30.09 % MAPE reduction compared to alternative expert architectures. Compared to the multi-gate mixture-of-experts based models, classical forecasting models, and recent advanced models, the proposed model achieves superior forecasting accuracy, validating its effectiveness and advancement.http://www.sciencedirect.com/science/article/pii/S266654682500093XIntegrated energy systemsLoad forecastingMulti-task learningMixed time scalesFeature fusion |
| spellingShingle | Yunfei Zhang Jun Shen Jian Li Mingzhe Yu Xu Chen Ziyong Yin Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention Energy and AI Integrated energy systems Load forecasting Multi-task learning Mixed time scales Feature fusion |
| title | Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention |
| title_full | Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention |
| title_fullStr | Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention |
| title_full_unstemmed | Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention |
| title_short | Achieving high precision and balanced multi-energy load forecasting with mixed time scales: a multi-task learning model with stacked cross-attention |
| title_sort | achieving high precision and balanced multi energy load forecasting with mixed time scales a multi task learning model with stacked cross attention |
| topic | Integrated energy systems Load forecasting Multi-task learning Mixed time scales Feature fusion |
| url | http://www.sciencedirect.com/science/article/pii/S266654682500093X |
| work_keys_str_mv | AT yunfeizhang achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention AT junshen achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention AT jianli achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention AT mingzheyu achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention AT xuchen achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention AT ziyongyin achievinghighprecisionandbalancedmultienergyloadforecastingwithmixedtimescalesamultitasklearningmodelwithstackedcrossattention |