摘要
为了解决风电随机波动带来的调频、调压和调度运行问题,需要准确预测风电场的出力。提出了一种新型短期风电功率预测方法。在对风电功率进行预测之前,对历史风电功率序列和气象数据进行预处理。首先采用集合经验模态分解将风电时间序列分解为若干分量,再采用混沌时间序列分析检测出混沌分量,接着用奇异谱分析消除混沌分量的高频低幅波动,通过消除风电功率序列的混沌性,以提高预测的准确性。然后引入局部线性嵌入算法对气象数据进行降维,以提高计算效率。在预测阶段,采用改进的极限学习机进行预测,通过灰狼优化算法优化极限学习机的输入权重和隐层偏置,并使用训练后的改进极限学习机预测短期风电功率。通过对南方电网某风电场的应用表明,该预测模型具有较好的预测精度。
In order to solve the problems of frequency modulation,voltage regulation and dispatching operation caused by wind power random fluctuation,it is necessary to accurately predict the output of wind farms.A new short-term wind power prediction method is proposed.Before predicting the wind power,the historical wind power series and meteorological data are preprocessed.Firstly,the time series of wind power is decomposed into several components by using ensemble empirical mode decomposition,then the chaotic components are detected by using the chaotic time series analysis.Then the high frequency and low amplitude fluctuations of the chaotic components are eliminated by using the singular spectrum analysis,so as to improve the accuracy of prediction by eliminating the chaos of the wind power series.Then the locally linear embedding algorithm is introduced to reduce the dimension of meteorological data,so as to improve the calculation efficiency.In the prediction stage,the improved extreme learning machine is used to predict.The input weight and hidden layer bias of the extreme learning machine are optimized by Grey Wolf Optimizer,and the short-term wind power is predicted by the improved extreme learning machine after training.The application of a wind farm in China Southern Power Grid shows that the prediction model proposed in this paper has better prediction accuracy.
作者
袁智勇
白浩
黄安迪
雷金勇
赵显秋
YUAN Zhiyong;BAI Hao;HUANG Andi;LEI Jinyong;ZHAO Xianqiu(Electric Power Research Institute of China Southern Power Grid,Guangzhou 510663,Guangdong,China;Department of Electrical Engineering,Southeast University,Nanjing 210096,Jiangsu,China)
出处
《水利水电技术（中英文）》
北大核心
2021年第S01期323-331,共9页
Water Resources and Hydropower Engineering
基金
中国南方电网有限责任公司科技项目“分散式风力发电高比例接入配电网关键技术研究”(ZBKJXM20180015)
关键词
短期风电功率预测
集合经验模态分解
混沌时间序列分析
奇异谱分析
局部线性嵌入
灰狼优化算法
极限学习机
short-term wind power prediction
ensemble empirical mode decomposition
chaotic time series analysis
singular spectrum analysis
locally linear embedding
Grey Wolf Optimizer
extreme learning machine
