基于声雷达技术与神经网络的风电预测研究

Prediction of Wind Power Based on Acoustic Radar Technology and Neural Network

针对传统的测风塔在风电场实际测风应用中存在的问题,以测风声雷达技术与BP神经网络为基础开展风电预测研究。首先,介绍了测风声雷达在风电产业多种场景应用中的优点;其次,提出了集成经验模态分解(ensemble empirical mode decomposition,EEMD)与神经网络的风电预测算法;最后,以安徽女儿岭风电场声雷达测风塔采样数据为例进行预测算法验证,分析了10、30、70、80、85、95 m处风速对应的BP神经网络最优隐含层神经元个数,并在此基础上验证了基于集成经验模态分解与神经网络预测算法在声雷达不同测风高度下风电预测的有效性。

Aiming at the problems existing in the application of traditional wind measuring tower in wind farm,this paper carries out wind power prediction research based on wind acoustic radar technology and BP neural network.Firstly,we introduce the advantages of wind detection radar in various scenarios of wind power industry;Secondly,the wind power prediction algorithm integrating empirical mode decomposition and neural network is proposed;Finally,taking the sampling data of the wind tower of the acoustic radar in Anhui Nüerling wind farm as an example,the optimal hidden layer number of BP neural network corresponding to the wind speed of 80 meters,85 meters and 95 meters is verified.On this basis,the effectiveness of wind power prediction is verified based on integrated empirical mode decomposition and neural network prediction algorithm under different wind measurement heights of acoustic radar.