针对多因素风速预测问题,提出一种基于变量相关注意力机制的短期风速预测模型。该模型通过序列对齐与变量选择模块筛选出与风速相关性强的因素序列并生成其他因素矩阵,采用变分模态分解算法将历史风速序列分解并进一步生成模态分量矩阵。将其他因素矩阵与模态分量矩阵拼接,得到因素矩阵。利用变量相关注意力机制表达因素矩阵与风速序列的相关关系,在此基础上实现风速预测。采用中国南方某风电场一年的实测数据验证了序列对齐与变量选择模块的作用、变分模态分解算法的分解效果以及变量相关注意力机制的特征提取能力。该模型在4个季节的平均绝对误差分别为0.17、0.17、0.13和0.14 m/s,明显优于其他对比模型,充分说明该模型在短期风速预测方面的优越性能。
Abstract
A variable correlation attention mechanism-based short-term wind speed forecasting model is proposed for the multi-factor wind speed forecasting issue. The sequences strongly correlated with the wind speed sequence are screened out based on the sequence alignment and variable selection module, then the matrix made up of the samples of those sequences is generated. Meanwhile, a mode matrix is constructed from the VMD-decomposed historical wind speed sequence. The two matrixes are concatenated to generate the factor matrix. After that, the correlativity between the factor matrix and wind speed sequence is expressed by the variable correlation attention mechanism, and used to forecast the future wind speeds. Effectiveness of the sequence alignment and variable selection module, VMD module and variable correlation attention mechanism is validated on a one-year wind speed data measured from a wind power plant in southern China. The proposed model achieves the MAE values of 0.17, 0.17, 0.13 and 0.14 m/s in spring, summer, autumn and winter, respectively. Its performance is significantly better than that of the other comparative models, demonstrating the superiority of the proposed model.
关键词
风力发电 /
风速 /
预测 /
变分模态分解 /
最大信息系数 /
Transformer模型 /
变量相关注意力机制
Key words
wind power /
wind speed /
forecasting /
variational mode decomposition /
maximal information coefficient /
Transformer model /
variable correlation attention mechanism
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 金吉, 王斌, 喻敏, 等. Lorenz方程优化EMD分解过程的短期风速预测[J]. 太阳能学报, 2021, 42(6): 342-348.
JIN J, WANG B, YU M, et al.Short-term wind speed prediction based on EMD optimized by Lorenz equation[J]. Acta energiae solaris sincia, 2021, 42(6): 342-348.
[2] 潘超, 李润宇, 蔡国伟, 等. 考虑风速属性约简聚类的组合预测模型[J]. 电网技术, 2022, 46(4): 1355-1362.
PAN C, LI R Y, CAI G W, et al.Combined forecasting model considering wind speed attribute reduction and clustering[J]. Power system technology, 2022, 46(4):1355-1362.
[3] 刘擘龙, 张宏立, 王聪, 等. 基于序列到序列和注意力机制的超短期风速预测[J]. 太阳能学报, 2021, 42(9):286-294.
LIU B L, ZHANG H L, WANG C, et al.Ultra-short-term wind speed prediction based on sequence-to- sequence and attention mechanism[J]. Acta energiae solaris sincia,2021, 42(9): 286-294.
[4] HU H L, WANG L, TAO R.Wind speed forecasting based on variational mode decomposition and improved echo state network[J]. Renewable energy, 2021, 164:729-751.
[5] JASEENA K U, KOVOOR B C.Decomposition-based hybrid wind speed forecasting model using deep bidirectional LSTM networks[J]. Energy conversion and management, 2021, 234: 113944.
[6] ERDEM E, SHI J.ARMA based approaches for forecasting the tuple of wind speed and direction[J]. Applied energy,2011, 88(4): 1405-1414.
[7] REN C, AN N,WANG J Z, et al.Optimal parameters selection for BP neural network based on particle swarm optimization:a case study of wind speed forecasting[J]. Knowledge-based systems, 2014, 56: 226-239.
[8] SANTAMARIA G, REYES-BALLESTEROS A,GERSHENSON C.Wind speed forecasting for wind farms:a method based on support vector regression[J]. Renewable energy, 2016, 85: 790-809.
[9] LI H M,WANG J Z, LU H Y,et al.Research and application of a combined model based on variable weight for short term wind speed forecasting[J]. Renewable energy, 2018, 116: 669-684.
[10] JIANG P, LI C.Research and application of an innovative combined model based on a modified optimization algorithm for wind speed forecasting[J]. Measurement,2018, 124: 395-412.
[11] 高桂革, 原阔, 曾宪文, 等. 基于改进CEEMD-CS-ELM的短期风速预测[J]. 太阳能学报, 2021, 42(7): 284-289.
GAO G G, YUAN K, ZENG X W, et al.Short-term wind speed prediction based on improved CEEMD-CS-ELM[J].Acta energiae solaris sincia, 2021, 42(7): 284-289.
[12] JIANG P, LIU Z K,NIU X S, et al.A combined forecasting system based on statistical method, artificial neural networks, and deep learning methods for short-term wind speed forecasting[J]. Energy, 2021, 217: 119361.
[13] ZHANG J L, WEI Y M, TAN Z F.An adaptive hybrid model for short term wind speed forecasting[J]. Energy,2020, 190: 115615.
[14] WANG X G,REN H, ZHAI J H, et al.Adaptive support segment based short-term wind speed forecasting[J]. Energy, 2022, 249: 123644.
[15] SHANG Z H,HE Z S, CHEN Y, et al.Short-term wind speed forecasting system based on multivariate time series and multi-objective optimization[J]. Energy, 2022,238(Part C): 122024.
[16] WU B R, WANG L, ZENG Y R.Interpretable wind speed prediction with multivariate time series and temporal fusion transformers[J]. Energy, 2022, 252: 123990.
[17] 凡航, 张雪敏, 梅生伟, 等. 基于时空神经网络的风电场超短期风速预测模型[J]. 电力系统自动化, 2021, 45(1): 28-35.
FAN H, ZHANG X M, MEI S W,et al.Ultra-short-term wind speed prediction model for wind farms based on spatiotemporal neural network[J]. Automation of electric power systems, 2021, 45(1): 28-35.
[18] ZHANG S, CHEN Y, XIAO J H, et al.Hybrid wind speed forecasting model based on multivariate data secondary decomposition approach and deep learning algorithm with attention mechanism[J]. Renewable energy, 2021, 174:688-704.
[19] RESHEF D N, RESHEF Y A, FINUCANE H K, et al.Detecting novel associations in large data sets[J]. Science,2011, 334(6062): 1518-1524.
[20] VASWANI A, SHAZEER N, PARMAR N, et al.Attention is all you need[C]//Proceedings of the 31st Conference on Neural Informational Conference on Neural Information Processing Systems(NIPS), California, USA, 2017.
[21] DRAGOMIRETSKIY K, ZOSSO D.Variational mode decomposition[J]. IEEE transactions on signal processing,2014, 62(3): 531-544.
[22] ZHANG G W, LIU D.Causal convolutional gated recurrent unit network with multiple decomposition methods for short-term wind speed forecasting[J]. Energy conversion and management, 2020, 226: 113500.
[23] 刘长良, 武英杰, 甄成刚. 基于变分模态分解和模糊C均值聚类的滚动轴承故障诊断[J]. 中国电机工程学报,2015, 35(13): 3358-3365.
LIU C L, WU Y J, ZHEN C G.Rolling bearing fault diagnosis based on variational mode decomposition and fuzzy C means clustering[J]. Proceedings of the CSEE,2015, 35(13): 3358-3365.
基金
国家自然科学基金（62076093）; 河北省机器学习与计算智能重点实验室开放基金（2020-2022-A）
PDF(2108 KB)
