基于EMD分解和集对分析的风电功率实时预测

风电功率时间序列的随机性和波动性使得风电功率多步预测时难以达到理想的预测准确度,因此,提出一种基于经验模态分解(EMD)和集对分析的风电功率实时预测模型。该模型首先将风电功率时间序列经EMD分解,处理成有限个相对平稳的分量;然后利用极值点划分法,按波动程度相近的原则将分量重构为高频、中频和低频3个分量;最后对3个分量各自的特点针对性地建立预测模型,并将3个分量的预测结果叠加作为原始风电功率的预测值,用滚动的方式实现多步预测。采用3个不同装机容量的风电场的实测风电功率数据进行仿真,结果表明该方法提高了多步预测的准确度,显示出了良好的预测性能。

Real-Time Prediction for Wind Power Based on EMD and Set Pair Analysis

The randomness and volatility of wind power time series make it difficult to achieve the desired multi-step prediction accuracy. Therefore, a model of real-time prediction for wind power based on empirical mode decomposition ( EMD ) and set pair analysis is presented. The proposed wind power sequences are firstly decomposed into a series of functions with more stationary variation by the EMD technique. Then these functions are divided into three components( high-middle-low frequency components) according to their run-lengths by the extreme point division method. Finally, three prediction models are built under the basis of their respective variation rules, and the results of three prediction models are reconstructed with the original wind power prediction value, this model achieves multi-step prediction by rolling prediction. The data from Three different wind farms with different installed capacity are used for simulate experiment. The results show that the proposed approach possesses with higher accuracy and the prediction performance is satisfied.