负荷记录中的噪声以及预测方法中矩阵数值计算的奇异性,使得一次预测得到的结果具有较大的误差。为了降低初值中噪声的不利影响,将数值天气预报中的Ensemble方法移植到短期负荷预测中。在混沌相空间重构预测中,在参考矢量上叠加一定强度的正态分布噪声,形成多个扰动后的参考矢量,分别预测后得到多个预测结果,再由这些预测结果合成概率化的预测结果。采用这种Ensemble技术,不仅可以提高预测准确率,还可以得到概率化的预测结果。
Measurement noises and the singularity in matrix numerical computation bring errors in short term load forecasting. To reduce the influence of noises on initial values, Ensemble prediction in numerical weather prediction is thereafter used in short term load forecasting. In chaotic phase space reconstruction forecasting, level noises with normal distribution are added to the reference vector to form many reference vectors with perturbations. Many forecasting results are respectively given based on the different reference vectors with perturbations to obtain the probabilistic forecasting result. The higher accuracy and probabilistic characteristic are two main advantages rendered by the Ensemble technique of chaotic forecasting.
[1] | 杨正瓴,王渭巍,曹东波,等.短期负荷预测的Ensemble混沌预测方法[J].电力系统自动化,2007,31(23):34-37. YANG Zhengling, WANG Weiwei, CAO Dongbo, et al. Ensemble Chaotic Forecasting Method in Short Term Load Forecasting[J]. Automation of Electric Power Systems, 2007, 31(23):34-37. |