基于信息熵的历史数据选取对超短期风电功率预测精度影响研究

Impact of Historical Data Selection on Accuracy of Ultra-short-term Wind Power Prediction Based on Prediction Information Entropy

风电功率超短期预测通常依据历史数据滚动预测得到,因此,历史数据的时间分辨率对预测结果的准确性有显著影响。首先给出"预测信息熵"指标,预测信息熵兼顾不同时间分辨率下数据序列的信息含量和超短期预测时预测步数2个因素,表征不同时间分辨率的功率序列在超短期风电功率预测时的精确预测能力,最后提出基于"预测信息熵"选择超短期风电功率预测时应采取的时间分辨率的方法。以某风电场实测数据为例,采用最小二乘支持向量机模型进行风电功率超短期预测,预测过程中历史数据分辨率的选取采用了所提的基于预测信息熵的选取方法,预测结果验证了所提方法的有效性。

The ultra-short term wind power prediction is usually obtained from the historical data,therefore,the time resolutions of the historical data has a significant impact on the accuracy of the results in the wind power prediction.A index of"prediction information entropy",which considers both the information content and the forecasting steps of the wind power series with different time resolutions,is proposed to represent the prediction performance of the wind power with different time resolutions in this paper.Then the method of choosing the best time resolutions based on the prediction information entropy is proposed.Finally,taking the measured data of a wind farm as an example,the ultra-short-term wind power prediction is carried out based on the least square support vector machine model.In the prediction process,the time resolution of the historical data is selected with the above mentioned method based on the prediction information entropy,and the prediction results verifies the effectiveness of the proposed method.