基于高阶Markov链模型的风电功率预测性能分析

Analysis of the wind power forecasting performance based on high-order Markov chain models

为了提高短期风电功率预测的精度,提出一种基于Markov链理论的预测算法。该算法直接对风电功率数据进行分析,划分了四种状态空间,并根据状态空间数和建模数据量的不同分别建立一阶和二阶Markov链模型。采用新误差公式NRMSE,给出不同状态空间数和建模数据量下的一阶、二阶Markov链模型预测性能比较结果。进一步给出在选取相同状态空间数、相同建模数据量的情况下,一阶和二阶Markov链模型的灵敏度分析。经实例验证,该算法能有效地提高单点值预测精度,并且给出了与预测值相关的概率分布结果。

A forecasting algorithm based on Markov chain theory is proposed to improve the precision of short-term wind power forecasting. The data of the wind power are analyzed directly and four kinds of state-spaces are formed. The order- 1 and order-2 models are built according to the number of state-space and the differences of modeling quantities. The comparison results between order-I and order-2 Markov models under different numbers of state-spaces and modeling data are presented through the new error formulaNRMSE. And then sensitivity analyses of order-1 and order-2 Markov models are provided based on the same number of state-spaces and modeling quantity. Experimental results show that the proposed method can improve the prediction accuracy, and it provides probability distribution results associated with prediction value.