.png){kind=icon}
返回
Quantum attention-based interpretable transfer learning for wind power probabilistic forecasting
.png){kind=icon}
PRE
AI
DOI：10.1016/j.engappai.2026.115783.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
To address the negative transfer problem caused by differences in variable types between the source and target domains during the prediction process for new wind farms, this study proposes a quantum attention-based interpretable transfer learning for wind power probabilistic forecasting (QAITL) method. The quantum attention mechanism integrated into QAITL can effectively capture temporal dependencies across variables, thereby enhancing model robustness against distribution shifts. Moreover, the QAITL achieves interpretable identification of critical variables via variable selection networks, which mitigates the impact of redundant features on predictive performance. To further address the negative transfer problem, the QAITL employs a two-stage transfer learning framework capable of the selective transfer of variable weights. This study conducts comprehensive experiments based on real wind farm datasets from three different geographical locations. In the zero-shot learning experiments, compared with the best-performing baseline model on each dataset, the QAMITL reduces MAE by 4.26%-6.42% and RMSE by 3.74%-5.02%. In the few-shot learning experiments, the QAMITL maintains stable performance improvement as the amount of target-domain data increases and avoids the negative transfer observed in the representative baseline model. Additionally, this study analyzes the interpretability of the proposed method, revealing the key variables affecting wind energy output in different datasets. Finally, this study verifies the role and significance of each core module in the proposed method through ablation experiments.
期刊
IF:
8
论文数:
5.8K
被引数:
3.5W
机构
引用论文
.png){kind=icon}
QFTD: An efficient quantum federated learning for transformer fault diagnosis with minimal gated unit in smart gridQFTD：一种针对智能电网变压器故障诊断的最小门控单元高效量子联邦学习方法
PREAI
分享收藏Physics-informed cross layer temporal frequency transformer network for remaining useful life prediction of rolling bearings融合物理信息的跨层时空频率变换器网络在滚动轴承剩余使用寿命预测中的应用
PREAI
分享收藏Denoising diffusion models with optimized quantum implicit neural networks for image generation用于图像生成的优化量子隐式神经网络去噪扩散模型
PREAI
分享收藏Investigating Personalization Techniques for Improved Cybersickness Prediction in Virtual Reality Environments研究虚拟现实环境中用于提高网络眩晕预测的个性化技术
PREAI
分享收藏Short-term regional wind power forecast method based on hierarchical graph neural network and extreme value theory基于分层图神经网络和极值理论的短期区域风电预测方法
PREAI
分享收藏A novel photovoltaic power probabilistic forecasting model based on monotonic quantile convolutional neural network and multi-objective optimization基于单调分位数卷积神经网络和多目标优化的光伏发电概率预测模型
PREAI
分享收藏
查看更多内容