摘要
随着风电渗透率的不断提高,对风电功率进行精准、可靠的预测是提升风电消纳水平的有效措施。针对功率预测时风电数据种类不足和特征数量稀缺的问题,提出基于混合特征双重衍生和误差修正的风电功率超短期预测模型。首先,在原始功率特征中施加混沌噪声,构造出多条混沌扰动特征,改善原始功率特征分布过于单一的状况。其次,提出基于免疫算法的特征衍生算法,挖掘风电功率数据的潜在信息,增加优质特征数量,进而构建误差预测模型,通过预测风电功率预测误差修正风电功率预测结果,进一步提升预测准确率。最后,基于比利时风电场实际运行数据进行算例分析。所提模型预测效果较好,且相较其他传统预测模型精确度更高,验证了所提模型的有效性。
With the increase of the wind power penetration rate,accurate and reliable wind power forecasting is an effective measure to improve the accomodation level of wind power.Addressing the issues of insufficient types of wind power data and scarcity of feature quantities in power forecasting,an ultra-short-term forecasting model of wind power based on the dual derivation of hybrid features and error correction is proposed.Firstly,the chaotic noise is applied to the original power features to construct multiple chaotic disturbance features,improving the situation where the distribution of the original power features is too simple.Secondly,a feature derivation algorithm based on immune algorithms is proposed to explore the potential information of wind power data and increase the number of high-quality features.Furthermore,an error forecasting model is constructed to correct the wind power forecasting results and further improve the forecasting accuracy by forecasting the wind power forecasting errors.Finally,based on the actual operation data of Belgian wind farms,a numerical case analysis is conducted.The proposed model shows better forecasting performance and higher accuracy compared with other traditional forecasting models,verifying the effectiveness of the proposed model.
作者
袁畅
王森
孙永辉
武云逸
谢东亮
YUAN Chang;WANG Sen;SUN Yonghui;WU Yunyi;XIE Dongliang(College of Energy and Electrical Engineering,Hohai University,Nanjing 210098,China;NARI Group Corporation(State Grid Electric Power Research Institute),Nanjing 211106,China)
出处
《电力系统自动化》
EI
CSCD
北大核心
2024年第5期68-76,共9页
Automation of Electric Power Systems
基金
国家自然科学基金资助项目(62073121)
南瑞集团有限公司项目“信息-物理-社会元素的交互及协调技术(GF-GFWD-210338)”资助。
关键词
风电功率预测
风电场
特征稀缺回归预测
特征衍生
误差修正
超短期预测
wind power forecasting
wind farm
feature scarcity regression forecasting
feature derivation
error correction
ultra-short-term forecasting
