摘要
光伏发电功率的准确预测对于提高电网和微电网的供电质量和降低运行成本具有重要意义。针对光伏发电时间序列的非线性和非平稳特征,在传统基于经验模态分解(EMD)方法的基础上,加入白噪声检验环节,提出一种基于改进EMD和差分自回归移动平均(ARIMA)相结合的光伏发电系统短期功率预测方法。首先利用EMD将原始光伏发电系统功率序列分解为多个具有不同频率的固有模态函数(IMF)分量,并对各IMF分量进行白噪声检验,筛选出不含白噪声的IMF分量;然后,对分量进行平稳性检验,对非平稳分量进行平稳化处理;最后对平稳的分量序列分别建立ARIMA预测模型,将各分量预测值进行叠加得到最终预测值。为验证方法的有效性,利用实证系统对3种天气条件下共15天的光伏发电功率进行了预测,并与传统的ARIMA、EMD-AR和EMD-ARIMA等方法进行了对比。误差统计结果表明,在相同样本数据量的前提下,该方法预测误差普遍低于其它方法。
The accurate prediction of photovoltaic power generation is of great significance to improve the power supply quality of the power grid and microgrid and reduce operating costs.Aiming at the non-linear and non-stationary characteristics of photovoltaic power generation time series,based on the traditional empirical mode decomposition(EMD)method,this study introduces the white noise test link and proposes a short-term power prediction method based on improved empirical mode decomposition(EMD)and auto regressive integrated moving average(ARIMA)model.Firstly,the original PV output sequence is decomposed into multiple Intrinsic Mode Function(IMF)components with different frequencies by EMD,and white noise test on each IMF component is performed to filter out IMF components without white noise.Then,the components are tested for stationarity,and those non-stationary components are smoothed.Finally,the stationary component sequences are reconstructed by ARIMA prediction models,respectively,and the final prediction result is obtained by superimposing prediction value of each component.In order to verify the effectiveness of the method,a demonstrated system is used to predict the PV output for a total of 15 days under three weather conditions and the result is compared with the three methods of traditional ARIMA,EMD-AR,and EMD-ARIMA.The error statistics show that the prediction error of this method is generally lower than the other three methods under the premise of the same sample data volume.
作者
仇琦
杨兰
丁旭
董志强
苏然
郑凌蔚
QIU Qi;YANG Lan;DING Xu;DONG Zhiqiang;SU Ran;ZHENG Lingwei(School of Automation,Hangzhou Dianzi University,Hangzhou 310018,China)
出处
《电力科学与工程》
2020年第8期42-50,共9页
Electric Power Science and Engineering
关键词
光伏发电系统功率预测
时间序列
经验模态分解
差分自回归移动平均
白噪声检验
平稳性检验
power output prediction of PV power system
time series
empirical mode decomposition(EMD)
auto regressive integrated moving average(ARIMA)
white noise test
stationarity test
