为提高风电功率概率预测精度和缩短长短期记忆网络的训练时间,提出一种基于分位数回归结合新遗忘门长短期记忆（NFGLSTM）网络与核密度估计的风电功率概率预测方法。该方法对长短期记忆网络的结构改进,提出一种新的遗忘门结构,以缩短训练时间。基于分位数回归和NFGLSTM网络建立组合预测模型,得到风电功率点预测值和某一置信度下的预测区间,采用Cosine核函数的核密度估计求解预测值的概率密度函数。基于某风电场的实测数据的算例分析表明,和传统预测方法相比,该方法可缩短长短期记忆网络的训练时间,提高概率预测精度。
Abstract
This paper proposes a wind power probability prediction method in order to improve the accuracy of wind power probability prediction and reduce the training time of long short-term memory network. This method is based on quantile regression combined with a new forget gate long short-term memory （NFGLSTM） network and kernel density estimation. The structure of long short-term memory network is improved and a new forget gate structure is proposed, which is used to shorten the training time. A combined forecasting model is established based on quantile regression and NGFLSTM network. So, the point prediction value of wind power and the prediction interval under a certain confidence are obtained. The kernel density estimation of the Cosine kernel function is used to solve probability density function of the predicted value. The case study shows that the proposed method can shorten the training time of long and short-term memory networks and improve the probability prediction accuracy.
关键词
风电功率 /
预测 /
长短期记忆 /
分位数回归 /
核密度估计
Key words
wind power /
prediction /
long short-term memory /
quantile regression /
kernel density estimation
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 薛禹胜, 雷兴, 薛峰, 等. 关于风电不确定性对电力系统影响的评述[J]. 中国电机工程学报, 2014, 34(29):5029-5040.
XUE Y S, LEI X, XUE F, et al.Comment on the impact of wind power uncertainty on power systems[J]. Proceedings of the CSEE, 2014, 34(29): 5029-5040.
[2] NIELSEN H A, MADSEN H, NIELSEN T S.Using quantile regression to extend an existing wind power forecasting system with probabilistic forecasts[J]. Wind energy, 2010, 9(1-2): 95-108.
[3] ZHANG C, WEI H K, ZHAO X, et al.A Gaussian process regression based hybrid approach for short-term wind speed prediction[J]. Energy conversion and management, 2016, 126: 1084-1092.
[4] WANG H Z, WANG G B, LI G Q, et al.Deep belief network based deterministic and probabilistic wind speed forecasting approach[J]. Applied energy, 2016, 182: 80-93.
[5] 杨锡运, 马雪, 张洋, 等. 基于EMD与加权马尔可夫链QR法的风电功率区间预测[J]. 太阳能学报, 2020, 41(2): 66-72.
YANG X Y, MA X, ZHANG Y, et al.Wind power interval prediction based on EMD and weighted Markov chain QR method[J]. Acta energiae solaris sinica, 2020, 41(2): 66-72.
[6] 杨锡运, 张艳峰, 叶天泽, 等. 基于朴素贝叶斯的风电功率组合概率区间预测[J]. 高电压技术, 2020, 46(3): 1099-1108.
YANG X Y, ZHANG Y F, YE T Z, et al.Probabilistic interval prediction of wind power combination based on naive Bayes[J]. High voltage technology, 2020, 46(3): 1099-1108.
[7] 郎伟明, 麻向津, 周博文, 等. 基于LSTM和非参数核密度估计的风电功率概率区间预测[J]. 智慧电力, 2020, 48(2): 31-37, 103.
LANG W M, MA X J, ZHOU B W, et al.Probability interval prediction of wind power based on LSTM and non-parametric kernel density estimation[J]. Smart electricity, 2020, 48(2): 31-37, 103.
[8] 钱勇生, 邵洁, 季欣欣, 等. 基于LSTM-Attention网络的短期风电功率预测[J]. 电机与控制应用, 2019, 46(9): 95-100.
QIAN Y S, SHAO J, JI X X, et al.Short-term wind power prediction based on LSTM-Attention network[J]. Motor and control applications, 2019, 46(9): 95-100.
[9] ZHANG Z D, YE L, QIN H, et al.Wind speed prediction method using shared weight long short-term memory network and Gaussian Process Regression[J]. Applied energy, 2019, 247(AUG. 1): 270-284.
[10] ZHANG Z D, QIN H, LU Y Q, et al.Wind speed forecasting based on quantile regression minimal gated memory network and kernel density estimation[J]. Energy conversion and management, 2019, 196(5): 1395-1409.
[11] PEI S Q, QIN H, ZHANG Z D, et al.Wind speed prediction method based on empirical wavelet transform and new cell update long short-term memory network[J]. Energy conversion and management, 2019, 196(SEP): 779-792.
[12] DUONG T, HAZELTON M L.Cross-validation bandwidth matrices for multivariate kernel density estimation[J]. Scandinavian journal of statistics, 2005, 32(3): 485-506.
[13] 李锦程. 基于门控循环单元的短期风电功率预测研究[D]. 徐州: 中国矿业大学, 2019.
LI J C.Research on short-term wind power prediction based on gated circulation unit[D]. Xuzhou: China University of Mining and Technology, 2019.
[14] QUINION R C, REIS E A, BESSEGATO L F.Using the coefficient of determination R2 to test the significance of multiple linear regression[J]. Teaching statistics, 2013, 35(2): 84-88.
基金
国家自然科学基金（51677121; 51537007）; 辽宁省教育厅服务地方项目（LFGD2017009）; 辽宁省“兴辽英才计划”项目（XLYC1802041）
PDF(1912 KB)
