基于EWT-WKELM的短期负荷预测  

Short-term Power Load Forecasting Based on EWT-WKELM

作　　者：李青 于永军 郑少鹏 马天娇 LI Qing;YU Yongjun;ZHENG Shaopeng;MA Tianjiao(Electric Power Research Institute,State Grid Xinjiang Electric Power Co.,Ltd,Urumqi 830000,China;Xinjiang Railway Vocational and Technical College,Urumqi 830000,China)

机构地区：[1]国网新疆电力公司电力科学研究院,乌鲁木齐830000 [2]新疆铁道职业技术学院,乌鲁木齐830000

出　　处：《电力系统及其自动化学报》2018年第7期83-89,共7页Proceedings of the CSU-EPSA

基　　金：基于串阻型逆变器的光伏电站并网特性实证性研究与测试(5230DK160006);电力视频大数据分布式计算及智能分析关键技术研究及应用(5230DK15003N)

摘　　要：为提高短期负荷预测的精度,提出一种基于经验小波变换EWT(empirical wavelet transform)和小波核极限学习机WKELM(wavelet kernel extreme learning machine)的组合预测方法。首先,采用EWT将乌鲁木齐地区的实测负荷原始序列分解为具有特征差异的不同分量;然后,采用小波核极限学习机对各分解负荷子序列分别进行预测;最后,叠加各分量预测值得到最终的预测结果。实验结果表明,相比WKELM单一预测方法,该方法可将平均绝对值百分比误差MAPE(mean absolute percentage error)降低82.8%;相比EMD-WKELM组合预测方法,该方法在大量降低组合预测规模的同时,仍可将MAPE降低69.8%。Based on empirical wavelet transform(EWT)and wavelet kernel extreme learning machine(WKELM),a combined forecasting method is proposed to improve the short-term power load forecasting accuracy.First,the original load sequences in Urumqi area were decomposed into different load components with different characteristics using EWT.Then,WKELM was used to forecast each decomposed component,respectively.Finally,the predictive value of each component was superimposed to form the final prediction results.Experimental results show that compared with the WKELM forecasting method,the proposed method can reduce the mean absolute percentage error(MAPE)by 82.8%;compared with the EMD-WKELM forecasting method,it still can reduce MAPE by 69.8%under the condition of a large reduction in the combined prediction scale.

关 键 词：负荷预测 经验小波变换 极限学习机 小波核函数 

分 类 号：TM715[电气工程—电力系统及自动化]