针对现有短期光伏功率区间预测问题,提出一种时间卷积神经网络与注意力机制结合的框架,对注意力机制中的时间因果顺序进行严格限制,应用残差机制增强模型挖掘的信息能力,并利用质量驱动区间损失优化模型参数,最终实现短期功率区间预测效果的提高。根据中国河北省某光伏电站的当地气象数据和历史光伏功率数据进行的仿真实验表明,相较于传统的序列预测方法或区间损失,在连续时刻和不同天气类型情况下,所提出的功率区间预测方法效果更有助于电网的科学调度与决策。
Abstract
For the existing problems of short-term photovoltaic power interval prediction, a framework combining a time convolution neural network with an attention mechanism is proposed. This framework imposes strict constraints on the temporal causal order in the attention mechanism, applies residual blocks to enhance the information mining ability of the model, and utilizes model parameters for quality-driven interval loss simultaneously, which ultimately improves the short-term power interval prediction effect. The simulation experiments based on the local meteorological data and historical photovoltaic power data of a photovoltaic power station in Hebei Province, China, show that compared with the traditional sequence prediction method or interval loss, the power interval prediction method proposed in this paper is more effective for scientific dispatching and decision-making of the power grid in continuous time and different weather types.
关键词
光伏发电 /
功率预测 /
深度学习 /
时间卷积网络 /
因果注意力机制 /
质量驱动损失
Key words
PV power /
power forecasting /
deep learning /
temporal convolutional network /
causal attention mechanism /
quality-driven loss
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 张成, 白建波, 兰康, 等. 基于数据挖掘和遗传小波神经网络的光伏电站发电量预测[J]. 太阳能学报, 2021, 42(3): 375-382.
ZHANG C, BAI J B, LAN K, et al.Photovoltaic power generation prediction based on data mining and genetic wavelet neural network[J]. Acta energiae solaris sinica, 2021, 42(3): 375-382.
[2] 万灿, 宋永华. 新能源电力系统概率预测理论与方法及其应用[J]. 电力系统自动化, 2021, 45(1): 2-16.
WAN C, SONG Y H.Theories, methodologies and applications of probabilistic forecasting for power systems with renewable energy sources[J]. Automation of electric power systems, 2021, 45(1): 2-16.
[3] WAN C, LIN J, SONG Y H, et al.Probabilistic forecasting of photovoltaic generation: an efficient statistical approach[J]. IEEE transactions on power systems, 2017, 32(3): 2471-2472.
[4] 朱晓荣, 金绘民, 王羽凝. 基于混合高斯模型与Copula函数结合的光伏电站功率相依结构建模[J]. 太阳能学报, 2019, 40(7): 1912-1919.
ZHU X R, JIN H M, WANG Y N.Analysis on dependence structure among photovoltaic power outputs based on combination of Copula function and Gaussian mixture model[J]. Acta energiae solaris sinica, 2019, 40(7): 1912-1919.
[5] KHOSRAVI A, NAHAVANDI S, CREIGHTON D, et al.Lower upper bound estimation method for construction of neural network-based prediction intervals[J]. IEEE transactions on neural networks, 2011, 22(3): 337-346.
[6] PEARCE T, ZAKI M, BRINTRUP A, et al.High-quality prediction intervals for deep learning: a distribution-free,ensembled approach[C]//International Conference on Machine Learning. Stockholm, Sweden, 2018.
[7] LIU H Y, HAN H, SUN Y, et al.Short-term wind power interval prediction method using VMD-RFG and Att-GRU[J]. Energy, 2022, 251: 123807.
[8] HU J M, ZHAO W G, TANG J W, et al.Integrating a softened multi-interval loss function into neural networks for wind power prediction[J]. Applied soft computing, 2021, 113: 108009.
[9] ALCÁNTARA A, GALVÁN I M, ALER R. Direct estimation of prediction intervals for solar and wind regional energy forecasting with deep neural networks[J]. Engineering applications of artificial intelligence, 2022, 114: 105128.
[10] 黄睿, 杜文娟, 王海风. 计及湍流强度的风电功率短期预测[J]. 电网技术, 2019, 43(6): 1907-1914.
HUANG R, DU W J, WANG H F.Short-term prediction of wind power considering turbulence intensity[J]. Power system technology, 2019, 43(6): 1907-1914.
[11] 李桂兰, 杨杰, 周满国. 基于改进时间卷积网络的光伏发电功率预测[J]. 激光与光电子学进展, 2022, 59(8): 480-489.
LI G L, YANG J, ZHOU M G.Power prediction of photovoltaic generation based on improved temporal convolutional network[J]. Laser & optoelectronics progress, 2022, 59(8): 480-489.
[12] MISHRA N, ROHANINEJAD M, CHEN X, et al.A simple neural attentive meta-learner[C]//International Conference on Learning Representations. Toulon, France, 2017.
[13] 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. Long Beach, California, USA, 2017: 6000-6010.
[14] 陆欣, 沈艳霞, 陈杰, 等. 考虑风力发电随机性的超短期风电功率区间预测研究[J]. 太阳能学报, 2017, 38(5): 1307-1315.
LU X, SHEN Y X, CHEN J, et al.Ultra-short-term wind power intervals prediction considering randomness of wind power generation[J]. Acta energiae solaris sinica, 2017, 38(5): 1307-1315.
[15] AKSAN E, HILLIGES O.STCN: stochastic temporal convolutional network[C]//International Conference on Learning Representations. New Orleans, USA, 2019.
[16] YU F, KOLTUN V.Multi-scale context aggregation by dilated convolutions[C]//International Conference on Learning Representations. San Juan, Puerto Rico, 2016.
[17] HE K M, ZHANG X Y, REN S Q, et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA, 2016: 770-778.
[18] YAO T C, WANG J, WU H Y, et al.A photovoltaic power output dataset: multi-source photovoltaic power output dataset with Python toolkit[J]. Solar energy, 2021, 230: 122-130.
基金
国家基金（61973226）; 山西省自然科学基金（201901D211083; 20210302123189）; 新型电力系统重点实验室项目（SKLD22KM22）
PDF(1857 KB)
