基于EMD的LS-SVM短期风电功率预测研究

Short-term LS-SVM Wind Power Prediction Algorithm Based on EMD

风电的随机波动在并网后可能影响电网安全,为保证电网安全,需对风机的输出功率做出准确预测并及时采取措施.本文设计一种基于经验模态分解的最小二乘支持向量机短期风电功率预测算法.采用经验模态分解法对某台风机的实际功率数据进行处理、分解,得到稳定功率分量;并采用基于最小二乘支持向量机的功率预测算法对每一分量数据进行预测处理;最后将分量功率预测值按一定规律叠加、重组,得到优化风电功率预测值.测试结果表明,该算法与其他算法比较,预测误差更小,预测精度更高.

Wind power generation is random and fluctuating.Through wind power prediction,reasonable scheduling strategies can be formulated.The paper proposes a short-term LS-SVM(Least Square Least Squares-Support Vector Machine)wind power prediction Algorithm.First,the data is decomposed by EMD(Empirical Mode Decomposition).Then the LS-SVM is used to do prediction processing by the data.Finally,the short-term power prediction value is obtained by data recombination.The test shows that the algorithm has high prediction accuracy.