基于多特征数据与混合模型短期风速预测研究

Short Term Wind Speed Prediction Based on Multi-feature Data and Mixed Model

为提高短期风速预测的准确性与可靠性,提出了一种考虑多特征数据的新型混合预测模型。模型基于Stacking算法集成自适应模糊神经网络、数据分组预测模型、随机森林回归模型,同时结合时变滤波器改进的模态分解、自适应噪声模态分解完成数据深度二次分解。首先,对多特征原始数据进行数据预处理得到多维子序列矩阵,计算子序列排列熵以此重构子序列矩阵;然后,利用Stacking算法集成混合模型对不同频域范围内的时间序列矩阵完成预测。通过与经典模型对比,表明本文提出的考虑多特征数据的混合模型预测精度和模型稳定性有较大优势。

In order to improve the accuracy and reliability of short-term wind speed prediction,a new hybrid prediction model considering multi feature data was proposed.The model was built based on Stacking integration algorithm integrating adaptive fuzzy neural network,data grouping prediction model and random forest regression model.At the same time,combined with the improved mode decomposition of time-varying filter and adaptive noise mode decomposition,the data depth secondary decomposition was completed.Firstly,the multidimensional subsequence matrix was obtained by preprocessing the original data with multiple features.Then,the subsequence array entropy was calculated to reconstruct the subsequence matrix.Finally,the mixed model integrated by Stacking integration algorithm was used to predict sequences in different frequency domains.The comparison with the classical model shows that the proposed hybrid model considering multi feature data has greater advantages in prediction accuracy and model stability.