基于风速误差校正和ALO-LSSVM的风电功率预测  

WIND POWER FORECASTING BASED ON WIND SPEED ERROR CORRETION AND ALO-LSSVM

作　　者：王斌 魏成伟[2] 谢丽蓉[1] 包洪印 张洁琼 买买提热依木·阿布力孜[1] Wang Bin;Wei Chengwei;Xie Lirong;Bao Hongyin;Zhang Jieqiong;Maimaitireyimu Abulizi(School of Electrical Engineering,Xinjiang University,Urumqi 830047,China;School of New Energy and Control,Xinjiang Engineering University,Urumqi 830000,China;CSIC of HaiWei(Xinjiang)New Energy Co.,Ltd.、Urumqi 830002,China;THE A Xinjiang New Energy Co.,Lid.,Urumqi 830011,China)

机构地区：[1]新疆大学电气工程学院,乌鲁木齐830047 [2]新疆工程学院新能源与控制学院,乌鲁木齐830000 [3]中船重工(海为)新能源有限公司,乌鲁木齐830002 [4]特变电工新疆新能源股份有限公司,乌鲁木齐830011

出　　处：《太阳能学报》2022年第1期58-63,共6页Acta Energiae Solaris Sinica

基　　金：国家自然科学基金(62163034);自治区区域协同创新专项-科技援疆计划(2018E02072);新疆高校科学基金重点项目(XJEDU20161017)。

摘　　要：风资源的随机波动性引起的相位滞后性问题,导致风电功率预测精度不高,尤其是风速变化较快时,滞后性引起的预测误差较大。考虑到风速波动与风功率变化密切相关,提出一种非参数核密度估计和数值天气预报(NWP)相结合的方法,并对预测风速误差进行校正,改善了预测风速的相位滞后性;然后将校正后的风速和风功率作为输入数据进行风电功率预测;采用蚁狮算法(ALO)优化最小二乘支持向量机(LSSVM)参数,从而建立基于风速误差校正和ALO-LSSVM组合的风电功率预测模型。算例结果表明,所提方法风功率预测精度更高。The problem of phase lag caused by the random fluctuation of wind resources lead to low wind power forecasting accuracy,especially when the wind speed changes rapidly,and the prediction error caused by hysteresis is large.Considering that wind speed fluctuation is closely related to wind power variation,a method combining non-parametric kernel density estimation and numerical weather prediction(NWP)is proposed,and the predicted wind speed error is corrected to improve the phase lag of predicted wind speed.The corrected wind speed and wind power are then used as input data for wind power forecasting.Using an ant lion optimizer(ALO)is proposed to optimize the parameters of least squares support vector machine(LSSVM).Thus,a combined super-short-term wind power forecasting model based on wind speed error correction and ALO-LSSVM is established.The results of the example show that the proposed method wind power forecasting accuracy is higher.

关 键 词：风电功率预测 最小二乘支持向量机 误差校正 蚁狮算法 非参数核密度估计 

分 类 号：TM614[电气工程—电力系统及自动化] TM71