基于融合模型的风电场输出功率短期预测方法  

Short term prediction method for output power of wind farm based on fusion model

作　　者：王凌云[1] 夏展鹏 许弘雷[2] 周璇卿 

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002 [2]华中科技大学能源与动力工程学院,湖北武汉430074

出　　处：《电源技术》2015年第10期2259-2262,共4页Chinese Journal of Power Sources

基　　金：国家自然科学基金(51407104);湖北省教育厅自然科学研究项目(Q20121305);国家"863"计划课题(2012AA050207);湖北国智恒三大电力科技有限公司研究生科研创新基金(HBGZH-201211)

摘　　要：风电场功率短期预测对并网风力发电系统的运行有着重要意义,在考虑风速、温度、海拔等影响风电功率的主要因素的基础上,为提高风电场短期输出功率的预测精度,提出基于风速与风电功率的融合预测模型。首先针对风电功率的直接预测,采用自回归时间序列和广义回归神经网络的组合模型来预测;然后再利用该组合模型预测风速,根据风速与风电功率的关系间接求出预测的风电功率;最后将前两种组合预测模型进行再次组合,得到融合预测模型。以吉林洮北风电场的短期功率预测为例,运用Matlab软件编程实现本文所提出的算法,验证模型的准确性与可行性,得到融合预测模型的预测相对误差为7.156%,可有效提高大型风电场输出功率的预测精度。Forecasting short term wind power was very significant for the operation of grid-connected wind power generation systems. On the basis of the consideration of wind speed, temperature, elevation and the other relevant factors which affect the wind power prediction, a fusion prediction model based on wind speed and wind power was proposed in order to improve the precision of the short term output wind power. Firstly, a combined prediction model of wind power was constructed by autoregressive time series and generalized regression neural network to forecast the wind power directly. Then, the same combined prediction model was used to forecast the wind speed and the wind power is computed indirectly according to the relationship between wind speed and power. Finally, the fusion prediction model could be obtained by combining the two previous model. For validating the precision and feasibility,all the algorithm presented was implemented by Matlab software with the practical operation data from Jilin Taobei wind farm. As a result, the error of the fusion prediction model was 7.156% and the prediction accuracy of output wind power for large scale wind farm could be improved effectively.

关 键 词：风速预测 功率预测 自回归时间序列 广义回归神经网络 融合预测模型 

分 类 号：TM614[电气工程—电力系统及自动化]