摘要
提高风电出力的预测精度,可以减轻风电并网带来的不利影响。利用径向基函数神经网络(RBF)建立风电出力预测模型,并通过正交二乘算法(OLS)对RBF神经网络进行初步训练,以确定网络结构及隐含层各节点中心。在OLS算法训练的网络基础上引入蛙跳算法(SFLA),进一步对隐含层基函数的宽度值进行优化以提高网络的泛化能力。实例预测表明,在相同的网络结构及隐含层中心下,基函数宽度值优化后的RBF神经网络模型预测精度得到了提升。
Increasing the forecasting accuracy of wind power can alleviate the negative influence caused by wind power integration. The paper utilizes radial basis function neural network （RBF） to establish the wind power forecasting model and uses the orthogonal least squares algorithm（OLS） to preli- minarily train the RBF neural network to determine the network＇s structure and central nodes in the hidden layer. In addition, the paper introduces Shuffled Frog Leaping Algorithm （SFLA） to optimize the width value of each radial basis function on the foundation of preliminary-trained network to further improve the network＇s generalization ability. The forecasting example shows that the forecasting accuracy of RBF neural network with the width value further optimized is improved with the same network＇s structure and central nodes in the hidden layer.
出处
《电网与清洁能源》
2013年第9期62-67,共6页
Power System and Clean Energy
基金
supported by Innovation Program for Young Talents in Science and Technology of Fujian Province(Number:2011J05124)
Natural Science Foundation of Fujian Province(Number:2013J01176)
