基于EMD与SVM的风电功率短期预测

由于风电存在着不确定性,风电功率预测对于接入大量风电的电力系统意义重大.为了提高风电功率的预测精度,本文建立了基于经验模式分解法（EMD）与支持向量机（SVM）的复合预测模型.考虑到风力机组的输出有很强的非线性,该模型首先将训练数据按风速大小分成高、中、低3组,然后对各组的风电功率样本序列进行经验模式分解,并建立各个频带分量的支持向量机预测模型,各模型的预测结果等权求和即得到最终的功率预测值.使用风电场现场采集数据的预测结果,验证了该方法的可行性和有效性.

Short time prediction of wind power based on EMD and SVM

As a clean renewable energy, wind power has a lot of advantage which is beyond fossil ener- gy. But there is uncertainty in wind power, so it is attached great significance to wind power predic- tion, especially for the power system which contains a large number of wind power. In order to raise the accuracy of wind power prediction, a compound prediction model which based on EMD（empirical mode decomposition） and SVM（support vector machine） was built. Considering the great nonlinearity of wind turbine generators outputs, the training data were divided into three groups on the basis of wind speed, and the wind power series were deeomtxxsed into several sequences in different band by EMD. By building different SVM prediction model, the decomposed sample series were predicted sep- arately, then the final results were got by adding every predicted result equal rights. The feasibility and effectiveness of this model were proved by the results of predicting the real data which were col- lected in wind farm.