基于LS-SVM和核密度估计的概率性风电功率预测

Probabilistic Wind Power Prediction Based on LS-SVM and Kernel Density Estimation

摘要:

近年来风电在我国发展迅猛,但风速的不稳定性和间歇性,使风电功率也具有同样的性质,这样的电功率注入会带来电力系统运行的不稳定,因此,风电功率的预测对风电并网及使用具有重要意义。鉴于此,开展风电功率的短期预测研究,利用LS-SVM对风电功率进行建模并实现确定性的短期预测,在此基础上使用非参数统计法对确定性预测模型的预测误差进行拟合获得其密度函数,计算各功率段的置信区间以得到概率性预测结果,从而提高风电功率预测结果的实用性和可靠性。与常用的自回归滑动平均模型和BP神经网络模型进行对比实验,证明本方法的性能及优势。

Abstract:

In recent years, wind power has developed rapidly in China, but the instability and intermittence of wind speed make the wind power have the same nature. Such power injection will bring instability to the operation of the power system. Therefore, prediction of wind power is of great significance to wind power network and utilization. In view of this, short-term prediction of wind power is focused based on LS-SVM, and thus the deterministic short-term prediction is achieved. In addition, the non-parametric statistical method is used to fit the prediction error of the deterministic prediction model to estimate its density function, and calculate the confidence interval of each power segment to obtain probabilistic prediction results. As a result, the practicability and reliability of wind power prediction results are improved. Compared with the commonly used autoregressive moving average model and BP neural network model, the performance and advantages of the method are proved.