基于Bagging神经网络集成的风功率预测

鉴于准确预测风功率对风电并网系统安全、稳定运行具有重要意义,提出了基于Bagging神经网络集成的风功率预测模型。先利用拉伊达(3σ)准则对数据进行预处理得到有效的风机数据,结合灰色关联度和Relief算法对数据进行特征提取;其次在Bagging集成学习中使用Bootstrap抽样,随机产生K个训练集并用自组织RBF神经网络(ErrCor-RBF)分别对风功率进行预测;最后叠加K个预测结果取均值得到最终预测结果。仿真结果表明,Bagging神经网络集成的风功率预测模型性能更好、预测精度较高。

Wind Power Prediction Based on Bagging Integrated Neural Network

In view of the importance of accurate prediction of wind power for the safe and stable operation of wind power grid connected system,a wind power prediction model is proposed based on the integration of Bagging neural network.Firstly,the effective data of wind turbine is obtained by preprocessing the data according to the criteria of Layda(3σ).And then the data is extracted by combining the gray correlation degree and relief algorithm.Secondly,Bootstrap sampling is used in bagging integrated learning to randomly generate Ktraining sets and predict the wind power respectively by using the self-organized RBF neural network(ErrCor-RBF).Finally,the final prediction results are obtained by superposition of the average value of K prediction results.The simulation results show that the wind power prediction model integrated with bagging neural network has better performance and higher prediction accuracy.