《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (11): 3616-3624.DOI: 10.11772/j.issn.1001-9081.2022111749
所属专题: 前沿与综合应用
收稿日期:
2022-11-24
修回日期:
2023-03-10
接受日期:
2023-03-17
发布日期:
2023-04-03
出版日期:
2023-11-10
通讯作者:
王晓英
作者简介:
李博录(1997—),男,甘肃天水人,硕士研究生,主要研究方向:人工智能、时空气象预测基金资助:
Bolu LI, Li WU, Xiaoying WANG(), Jianqiang HUANG, Tengfei CAO
Received:
2022-11-24
Revised:
2023-03-10
Accepted:
2023-03-17
Online:
2023-04-03
Published:
2023-11-10
Contact:
Xiaoying WANG
About author:
LI Bolu, born in 1997, M. S. candidate. His research interests include artificial intelligence, spatio-temporal meteorological prediction.Supported by:
摘要:
时空序列预测任务在交通、气象、智慧城市等领域有着广泛应用。站点风速预测作为气象预测中的主要任务之一,需要结合降水、气温等外部因素,学习不同数据的时空特征。气象站点的不规则分布和风本身的固有间歇性成为实现高精度风速预测的挑战。为考虑多站点空间分布对风速的影响以获得准确可靠的预测结果,提出一种基于图的动态转换注意力网络(Graph-DSAN)风速预测模型。首先,利用不同站点之间的距离重新构建它们的连接;其次,使用局部采样的过程建模不同采样大小的邻接矩阵,实现图卷积过程中邻居节点信息的聚合与传递;接着,将时空位置编码(STPE)处理后的图卷积结果加入动态注意力编码器(DAE)和转换注意力解码器(SAD)以实现动态注意力计算,从而提取时空相关性;最后,利用自回归的方式形成多步预测。在纽约州15个站点的风速预测实验中,将所设计模型与ConvLSTM、图多注意力网络(GMAN)、时空图卷积网络(STGCN)、动态转换注意力网络(DSAN)和时空动态网络(STDN)进行比较,Graph-DSAN的12 h预测均方根误差(RMSE)分别降低了28.2%、6.9%、27.7%、14.4%和8.9%,验证了Graph-DSAN风速预测的准确性。
中图分类号:
李博录, 吴利, 王晓英, 黄建强, 曹腾飞. 基于图动态注意力网络的多站点风速预测[J]. 计算机应用, 2023, 43(11): 3616-3624.
Bolu LI, Li WU, Xiaoying WANG, Jianqiang HUANG, Tengfei CAO. Multi-site wind speed prediction based on graph dynamic attention network[J]. Journal of Computer Applications, 2023, 43(11): 3616-3624.
网络模块 | 模块层数 | |||||
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | 6 | |
n_Decoder | 1.38 | 1.37 | 1.23 | 1.40 | 1.55 | 1.57 |
n_Encoder | 1.51 | 1.48 | 1.23 | 3.48 | 2.80 | 1.66 |
n_Layer | 1.46 | 1.30 | 1.23 | 1.42 | 1.34 | 1.70 |
网络模块 | 注意力头数 | |||||
1 | 2 | 4 | 8 | 16 | 32 | |
n_Head | 1.43 | 1.33 | 1.34 | 1.23 | 1.31 | 1.37 |
表1 不同条件下的RMSE
Tab. 1 RMSE under different conditions
网络模块 | 模块层数 | |||||
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | 6 | |
n_Decoder | 1.38 | 1.37 | 1.23 | 1.40 | 1.55 | 1.57 |
n_Encoder | 1.51 | 1.48 | 1.23 | 3.48 | 2.80 | 1.66 |
n_Layer | 1.46 | 1.30 | 1.23 | 1.42 | 1.34 | 1.70 |
网络模块 | 注意力头数 | |||||
1 | 2 | 4 | 8 | 16 | 32 | |
n_Head | 1.43 | 1.33 | 1.34 | 1.23 | 1.31 | 1.37 |
时长/h | 模型 | RMSE | MAE | 时长/h | 模型 | RMSE | MAE | 时长/h | 模型 | RMSE | MAE |
---|---|---|---|---|---|---|---|---|---|---|---|
1 | ConvLSTM | 1.73 | 1.36 | 6 | ConvLSTM | 2.18 | 1.71 | 12 | ConvLSTM | 2.97 | 2.31 |
STDN | 1.70 | 1.31 | STDN | 2.12 | 1.70 | STDN | 2.34 | 1.76 | |||
GMAN | 1.74 | 1.38 | GMAN | 2.08 | 1.57 | GMAN | 2.29 | 1.75 | |||
STGCN | 1.50 | 1.16 | STGCN | 2.13 | 1.70 | STGCN | 2.95 | 2.30 | |||
Xgboost | 1.36 | 1.04 | Xgboost | 1.93 | 1.48 | Xgboost | 2.19 | 1.72 | |||
DSAN | 1.28 | 0.95 | DSAN | 2.07 | 1.52 | DSAN | 2.49 | 1.97 | |||
Graph⁃DSAN | 1.23 | 0.89 | Graph⁃DSAN | 1.85 | 1.43 | Graph⁃DSAN | 2.13 | 1.71 |
表2 不同模型多步风速预测结果的比较
Tab. 2 Comparison of multi-step wind speed prediction results of different models
时长/h | 模型 | RMSE | MAE | 时长/h | 模型 | RMSE | MAE | 时长/h | 模型 | RMSE | MAE |
---|---|---|---|---|---|---|---|---|---|---|---|
1 | ConvLSTM | 1.73 | 1.36 | 6 | ConvLSTM | 2.18 | 1.71 | 12 | ConvLSTM | 2.97 | 2.31 |
STDN | 1.70 | 1.31 | STDN | 2.12 | 1.70 | STDN | 2.34 | 1.76 | |||
GMAN | 1.74 | 1.38 | GMAN | 2.08 | 1.57 | GMAN | 2.29 | 1.75 | |||
STGCN | 1.50 | 1.16 | STGCN | 2.13 | 1.70 | STGCN | 2.95 | 2.30 | |||
Xgboost | 1.36 | 1.04 | Xgboost | 1.93 | 1.48 | Xgboost | 2.19 | 1.72 | |||
DSAN | 1.28 | 0.95 | DSAN | 2.07 | 1.52 | DSAN | 2.49 | 1.97 | |||
Graph⁃DSAN | 1.23 | 0.89 | Graph⁃DSAN | 1.85 | 1.43 | Graph⁃DSAN | 2.13 | 1.71 |
方法 | 预测时长为1 h | 预测时长为12 h | ||
---|---|---|---|---|
RMSE | MAE | RMSE | MAE | |
DSAN-NS | 1.31 | 0.98 | 2.42 | 1.92 |
DSAN-SS | 1.41 | 1.06 | 2.60 | 2.00 |
DSAN-NE | 1.48 | 1.10 | 2.45 | 1.97 |
DSAN-ND | 1.42 | 1.07 | 2.74 | 2.15 |
Graph⁃DSAN | 1.23 | 0.89 | 2.13 | 1.71 |
表3 不同模块对模型的影响
Tab. 3 Influence of different modules on model
方法 | 预测时长为1 h | 预测时长为12 h | ||
---|---|---|---|---|
RMSE | MAE | RMSE | MAE | |
DSAN-NS | 1.31 | 0.98 | 2.42 | 1.92 |
DSAN-SS | 1.41 | 1.06 | 2.60 | 2.00 |
DSAN-NE | 1.48 | 1.10 | 2.45 | 1.97 |
DSAN-ND | 1.42 | 1.07 | 2.74 | 2.15 |
Graph⁃DSAN | 1.23 | 0.89 | 2.13 | 1.71 |
1 | 刘苇航,叶涛,史培军,等. 气候变化对粮食生产风险的影响研究进展[J]. 自然灾害学报, 2022, 31(4):1-11. |
LIU W H, YE T, SHI P J, et al. Advances in the study of climate change impact on crop producing risk[J]. Journal of Natural Disasters, 2022, 31(4): 1-11. | |
2 | 孟鑫禹,王睿涵,张喜平,等. 基于经验模态分解与多分支神经网络的超短期风功率预测[J]. 计算机应用, 2021, 41(1): 237-242. 10.11772/j.issn.1001-9081.2020060930 |
MENG X Y, WANG R H, ZHANG X P, et al. Ultra-short-term wind power prediction based on empirical mode decomposition and multi-branch neural network[J]. Journal of Computer Applications, 2021, 41(1): 237-242. 10.11772/j.issn.1001-9081.2020060930 | |
3 | PIOTROWSKI P, BACZYŃSKI D, KOPYT M, et al. Analysis of forecasted meteorological data (NWP) for efficient spatial forecasting of wind power generation[J]. Electric Power Systems Research, 2019, 175: No.105891. 10.1016/j.epsr.2019.105891 |
4 | AGRAWAL S, BARRINGTON L, BROMBERG C, et al. Machine learning for precipitation nowcasting from radar images[EB/OL]. (2019-12-11) [2022-11-08].. |
5 | 孙丽华,严军峰,徐健锋. 基于多机器学习竞争策略的短时雷电预报[J]. 计算机应用, 2016, 36(9):2555-2559. 10.11772/j.issn.1001-9081.2016.09.2555 |
SUN L H, YAN J F, XU J F. Short-term lightning prediction based on multi-machine learning competitive strategy[J]. Journal of Computer Applications, 2016, 36(9): 2555-2559. 10.11772/j.issn.1001-9081.2016.09.2555 | |
6 | RAVURI S, LENC K, WILLSON M, et al. Skilful precipitation nowcasting using deep generative models of radar[J]. Nature, 2021, 597(7878): 672-677. 10.1038/s41586-021-03854-z |
7 | 石峰,楼文高,张博. 基于灰狼群智能最优化的神经网络PM2.5浓度预测[J]. 计算机应用, 2017, 37(10):2854-2860. 10.11772/j.issn.1001-9081.2017.10.2854 |
SHI F, LOU W G, ZHANG B. Neural network model for PM2.5 concentration prediction by grey wolf optimizer algorithm[J]. Journal of Computer Applications, 2017, 37(10): 2854-2860. 10.11772/j.issn.1001-9081.2017.10.2854 | |
8 | 王军,费凯,程勇. 基于改进Adaboost-BP模型在降水中的预测[J]. 计算机应用, 2017, 37(9):2689-2693. 10.11772/j.issn.1001-9081.2017.09.2689 |
WANG J, FEI K, CHENG Y. Prediction of rainfall based on improved Adaboost-BP model[J]. Journal of Computer Applications, 2017, 37(9): 2689-2693. 10.11772/j.issn.1001-9081.2017.09.2689 | |
9 | SCHER S. Toward data‐driven weather and climate forecasting: approximating a simple general circulation model with deep learning[J]. Geophysical Research Letters, 2018, 45(22): 12616-12622. 10.1029/2018gl080704 |
10 | SALMAN A G, KANIGORO B, HERYADI Y. Weather forecasting using deep learning techniques[C]// Proceedings of the 2015 International Conference on Advanced Computer Science and Information Systems. Piscataway: IEEE, 2015: 281-285. 10.1109/icacsis.2015.7415154 |
11 | MEHRKANOON S. Deep shared representation learning for weather elements forecasting[J]. Knowledge-Based Systems, 2019, 179: 120-128. 10.1016/j.knosys.2019.05.009 |
12 | TREBING K, MEHRKANOON S. Wind speed prediction using multidimensional convolutional neural networks[C]// Proceedings of the 2020 IEEE Symposium Series on Computational Intelligence. Piscataway: IEEE, 2020: 713-720. 10.1109/ssci47803.2020.9308323 |
13 | ZHANG S, TONG H, XU J, et al. Graph convolutional networks: a comprehensive review[J]. Computational Social Networks, 2019, 6: No.11. 10.1186/s40649-019-0069-y |
14 | STAŃCZYK T, MEHRKANOON S. Deep graph convolutional networks for wind speed prediction[C/OL]// Proceedings of the 29th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning [2022-11-08].. 10.14428/esann/2021.es2021-25 |
15 | YU B, YIN H, ZHU Z. Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting[C]// Proceedings of the 27th International Joint Conference on Artificial Intelligence. California: ijcai.org, 2018: 3634-3640. 10.24963/ijcai.2018/505 |
16 | LIN H, BAI R, JIA W, et al. Preserving dynamic attention for long-term spatial-temporal prediction[C]// Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2020: 36-46. 10.1145/3394486.3403046 |
17 | HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. 10.1162/neco.1997.9.8.1735 |
18 | SHI X, CHEN Z, WANG H, et al. Convolutional LSTM network: a machine learning approach for precipitation nowcasting[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems — Volume 1. Cambridge: MIT Press, 2015: 802-810. |
19 | WANG Y, LONG M, WANG J, et al. PredRNN: recurrent neural networks for predictive learning using spatiotemporal LSTMs[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2017: 879-888. |
20 | TREBING K, STAǸCZYK T, MEHRKANOON S. SmaAt-UNet: precipitation nowcasting using a small attention-UNet architecture[J]. Pattern Recognition Letters, 2021, 145: 178-186. 10.1016/j.patrec.2021.01.036 |
21 | RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[C]// Proceedings of the 2015 International Conference on Medical Image Computing and Computer-Assisted Intervention, LNCS 9351. Cham: Springer, 2015: 234-241. |
22 | 廖雪超,伍杰平,陈才圣. 结合注意力机制与LSTM的短期风电功率预测模型[J]. 计算机工程, 2022, 48(9):286-297, 304. 10.19678/j.issn.1000-3428.0062059 |
LIAO X C, WU J P, CHEN C S. Short-term wind power prediction model combining attention mechanism and LSTM[J]. Computer Engineering, 2022, 48(9): 286-297, 304. 10.19678/j.issn.1000-3428.0062059 | |
23 | WILSON Y, TAN P N, LUO L. A low rank weighted graph convolutional approach to weather prediction[C]// Proceedings of the 2018 IEEE International Conference on Data Mining. Piscataway: IEEE, 2018: 627-636. 10.1109/icdm.2018.00078 |
24 | SEO S, MOHEGH A, BAN-WEISS G. with recurrent neural networks for spatiotemporal forecasting[C]// Proceedings of the 7th International Workshop on Climate Informatics. Boulder, CO: National Center for Atmospheric Research in Boulder, 2017: 85-88. |
25 | 祁柏林,郭昆鹏,杨彬,等. 基于GCN-LSTM的空气质量预测[J]. 计算机系统应用, 2021, 30(3): 208-213. 10.15888/j.cnki.csa.007815 |
QI B L, GUO K P, YANG B, et al. Air quality prediction based on GCN-LSTM[J]. Computer Systems and Applications, 2021, 30(3): 208-213. 10.15888/j.cnki.csa.007815 | |
26 | XU K, HU W, LESKOVEC J, et al. How powerful are graph neural networks?[EB/OL]. (2019-02-22) [2022-11-11].. |
27 | KIPF T N, WEWLLING M. Semi-supervised classification with graph convolutional networks[EB/OL]. (2017-02-22) [2022-11-11].. 10.48550/arXiv.1609.02907 |
28 | GAO H, WANG Z, JI S. Large-scale learnable graph convolutional networks[C]// Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2018: 1416-1424. 10.1145/3219819.3219947 |
29 | VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2017: 6000-6010. |
30 | VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al. Graph attention networks[EB/OL]. (2018-02-04) [2021-11-12].. |
31 | BRONSTEIN M M, BRUNA J, LeCUN Y, et al. Geometric deep learning: going beyond Euclidean data[J]. IEEE Signal Processing Magazine, 2017, 34(4): 18-42. 10.1109/msp.2017.2693418 |
32 | ZHOU J, CUI G, HU S, et al. Graph neural networks: a review of methods and applications[J]. AI Open, 2020, 1: 57-81. 10.1016/j.aiopen.2021.01.001 |
33 | NIEPERT M, AHMED M, KUTZKOV K. Learning convolutional neural networks for graphs[C]// Proceedings of the 33rd International Conference on Machine Learning. New York: JMLR.org, 2016: 2014-2023. |
34 | BRUNA J, ZAREMBA W, SZLAM A, et al. Spectral networks and locally connected networks on graphs[EB/OL]. (2014-03-21) [2022-11-12].. 10.1017/cbo9780511761942.003 |
35 | DEFFERRARD M, BRESSON X, VANDERGHEYNST P. Convolutional neural networks on graphs with fast localized spectral filtering[C]// Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2016: 3844-3852. |
36 | Meteorological Development Laboratory/Office of Science and Technology/National Weather Service/NOAA/U.S. Department of Commerce. TDL U.S. and Canada surface hourly observations[DB/OL]. [2022-11-12].. |
37 | YAO H, TANG X, WEI H, et al. Revisiting spatial-temporal similarity: a deep learning framework for traffic prediction[C]// AAAI’19/IAAI’19/EAAI’19: Proceedings of the 33rd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2019: 5668-5675. 10.1609/aaai.v33i01.33015668 |
38 | ZHENG C, FAN X, WANG C, et al. GMAN: a graph multi-attention network for traffic prediction[C]// Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2020: 1234-1241. 10.1609/aaai.v34i01.5477 |
[1] | 薛桂香, 王辉, 周卫峰, 刘瑜, 李岩. 基于知识图谱和时空扩散图卷积网络的港口交通流量预测[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2952-2957. |
[2] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[3] | 李力铤, 华蓓, 贺若舟, 徐况. 基于解耦注意力机制的多变量时序预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2732-2738. |
[4] | 赵志强, 马培红, 黑新宏. 基于双重注意力机制的人群计数方法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2886-2892. |
[5] | 庞川林, 唐睿, 张睿智, 刘川, 刘佳, 岳士博. D2D通信系统中基于图卷积网络的分布式功率控制算法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2855-2862. |
[6] | 张春雪, 仇丽青, 孙承爱, 荆彩霞. 基于两阶段动态兴趣识别的购买行为预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2365-2371. |
[7] | 薛凯鹏, 徐涛, 廖春节. 融合自监督和多层交叉注意力的多模态情感分析网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2387-2392. |
[8] | 汪雨晴, 朱广丽, 段文杰, 李书羽, 周若彤. 基于交互注意力机制的心理咨询文本情感分类模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2393-2399. |
[9] | 高鹏淇, 黄鹤鸣, 樊永红. 融合坐标与多头注意力机制的交互语音情感识别[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2400-2406. |
[10] | 刘禹含, 吉根林, 张红苹. 基于骨架图与混合注意力的视频行人异常检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2551-2557. |
[11] | 李钟华, 白云起, 王雪津, 黄雷雷, 林初俊, 廖诗宇. 基于图像增强的低照度人脸检测[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2588-2594. |
[12] | 莫尚斌, 王文君, 董凌, 高盛祥, 余正涛. 基于多路信息聚合协同解码的单通道语音增强[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2611-2617. |
[13] | 熊武, 曹从军, 宋雪芳, 邵云龙, 王旭升. 基于多尺度混合域注意力机制的笔迹鉴别方法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2225-2232. |
[14] | 李欢欢, 黄添强, 丁雪梅, 罗海峰, 黄丽清. 基于多尺度时空图卷积网络的交通出行需求预测[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2065-2072. |
[15] | 毛典辉, 李学博, 刘峻岭, 张登辉, 颜文婧. 基于并行异构图和序列注意力机制的中文实体关系抽取模型[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2018-2025. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||