基于GRNN全信息神经网络的超短期风速预测研究  

Research on Ultra Short-Term Wind Speed Prediction Based on GRNN All Information Neural Network

作　　者：高阳[1] 钟宏宇[1,2] 葛延峰[3] 董存[4] 许傲然[1] 

机构地区：[1]沈阳工程学院电力学院,辽宁沈阳110136 [2]国网吉林通化供电公司,吉林通化134000 [3]国网辽宁省电力有限公司调控中心,辽宁沈阳110136 [4]国家电网公司调控中心,北京100031

出　　处：《测控技术》2016年第4期149-152,共4页Measurement & Control Technology

基　　金：辽宁省高等学校优秀人才支持计划(LJQ2014136)

摘　　要：为了更好地研究风功率预测,风速预测显得至关重要。国内神经网络文献均只表现出了短期风速预测,而对于超短期风速预测的神经网络数学模型却相对稀少。引入了GRNN神经网络,详细说明了该方法的超短期风速预测原理并建立了数学模型;为了使超短期风速预测精度有一个良好的对比性分析,将影响风电输出功率的各NWP(numerical weather prediection)信息(包括风速、风向、气温、气压)进行组合,以国内某风电场2014年5月份的各NWP数据进行算例分析,实验结果表明,GRNN全信息神经网络可以达到很好的预测精度,而且运算网络的稳定性甚优。In order to study the wind power forecasting better,the wind speed predicting is particularly important.Domestic neural network literaturse only show short-term Wind speed forecasting,while the neural network model is relatively rare in the ultra short term wind speed forecasting.Introduced GRNN,ultra short term wind speed forecasting principle of the method is decribed and mathematical model is established.In order to make the ultra short term wind speed forecasting accuracy has a good,contrast analysis,NWP informations（including wind speed,wind direction,temperature,and barometric pressure） affecting the output power of wind power are combined.Take the NWP data of a wind farm in May 2014 in China as example to analyze,experimental results show that full information GRNN can achieve a good prediction accuracy and network stability is very excellent.

关 键 词：GRNN 非线性 风速预测 超短期 RBF 

分 类 号：TM743[电气工程—电力系统及自动化]