| 引用本文: | 伍 耘,葛佳敏,王文烨,等.贝叶斯优化超参数的空时融合压缩残差网络在风速区间预测中的研究[J].电力系统保护与控制,2025,53(1):13-23.[点击复制] |
| WU Yun,GE Jiamin,WANG Wenye,et al.Wind speed interval prediction using spatio-temporal fusion compressed residual networks with Bayesian optimized hyperparameters[J].Power System Protection and Control,2025,53(1):13-23[点击复制] |
|
| 摘要: |
| 针对风电场规划中风速的高随机性问题，提出了一种基于小样本空时融合压缩残差网络点预测(spatio-temporal integration and compression deep residual, STiCDRS)模型。该模型旨在深入挖掘风速序列中的空间和时间特征，以提升点预测精度。首先，采用空时融合压缩残差网络点预测模型得到点预测结果。然后，在此基础上采用新颖的空时融合压缩残差网络区间(STiCDRS-Gaussian process regression, STiCDRS-GPR)预测模型得到风速的区间预测结果，进而得到更为可靠的风速概率预测结果。该模型采用贝叶斯优化方法进行超参数选择，确保超参数的高效自动化调优。最后，使用内蒙古地区风电场的风速数据集，将STiCDRS模型与传统经典模型的预测结果进行对比。实验结果表明，相较于其他模型，所提STiCDRS-GPR模型在风速预测中具有更高的点预测精度、适宜的预测区间以及可靠的概率预测结果，充分展示了其在风速预测领域的良好应用潜力。 |
| 关键词: 风速预测 时序卷积网络 STiCDRS模型 GPR区间预测 贝叶斯优化 |
| DOI：10.19783/j.cnki.pspc.240596 |
| 投稿时间：2024-05-14修订日期：2024-10-15 |
| 基金项目:国家重点研发计划项目资助(2021YFB2601504) |
|
| Wind speed interval prediction using spatio-temporal fusion compressed residual networks with Bayesian optimized hyperparameters |
| WU Yun1,GE Jiamin1,WANG Wenye1,LI Xiaoyong2,CHE Liang1 |
| (1. College of Electrical and Information Engineering, Hunan University, Changsha 410082, China;
2. School of Environment, South China Normal University, Guangzhou 510630, China) |
| Abstract: |
| To address the challenge of high wind speed variability in wind farm planning, a small-sample-based spatio- temporal integration and compression deep residual point prediction model, spatio-temporal integration and compression deep residual (STiCDRS), is proposed. This model is designed to deeply explore the spatial and temporal characteristics within wind speed sequences to enhance the accuracy of point prediction. Initially, the spatio-temporal integration and compression deep residual network is employed to obtain point prediction results. Subsequently, an innovative hybrid model, STiCDRS-Gaussian process regression (STiCDRS-GPR), is introduced to achieve interval prediction results, thereby providing more reliable probabilistic forecasts of wind speed. The model utilizes a Bayesian optimization method for hyperparameter selection, ensuring efficient and automated tuning. Finally, the wind speed dataset from a wind farm in Inner Mongolia is used to compare the prediction results of the STiCDRS model with those of traditional classical models. Experimental results demonstrate that, in comparison to other models, the proposed STiCDRS-GPR model delivers superior point prediction accuracy, appropriate interval predictions, and reliable probabilistic forecasting outcomes, fully showcasing its considerable potential in the domain of wind speed forecasting. |
| Key words: wind speed prediction temporal convolutional network STiCDRS model GPR interval prediction Bayesian optimization |
