摘要
为了提高风电功率的预测精度,基于多模型的预测MS-RBF神经网络的进行组合,通过Bayesian分类训练各子模型的权值,然后根据权重计算最终预测值;基于新疆某风电场实测历史数据,采用该组合模型与RBFNN模型分析对比,验证结果表明该组合模型有效减少了较大误差出现的频率,提高了整体的预测精度。
Increased the prediction accuracy of short-term wind turbine output power plays a key role in improving the electric grid system economic and security. In order to enhance the prediction accuracy of wind power, this paper selected the mixed structure-radial basis function （MS-RBF） neural network multi-model combined with each other, training the web weights by Bayesian learning, then reconstruction the final prediction for very short-term wind turbine output power; Based on the measured historical data of wind farm in Xinjiang, the compositional model compared with RBFNN models, experimental results indicated that the combination model effectively reduces the frequency of error and improve the overall prediction accuracy.
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2014年第3期457-461,共5页
Acta Energiae Solaris Sinica
