基于有偏厚尾ARMAX模型的短期电价预测

Short-term Electricity Price Forecasting Based on ARMAX Model with Skewness and Kurtosis

基于正态分布假设的时间序列分析模型不能有效地处理电价的有偏厚尾特征。在对电力市场现货电价的影响因素和波动规律综合分析的基础上,提出了一种基于有偏厚尾ARMAX模型的的短期电价预测方法。该方法采用正弦函数和基于正态分布概率密度函数的Gram-Charlier展开来描述多重周期性和有偏厚尾性,可同时考虑电价分布的多重周期性、有偏厚尾性及其与负荷之间的非线性相关性。对PJM电力市场历史数据的算例研究表明,该方法计算量小,待估参数少,能准确反映电价的变化规律,具有一定的实用价值。

The model of time series analysis with normal distribution can not effectively deal with the skewness and kurtosis of electricity spot price.With comprehensive consideration of the various influencing factors and the fluctuation rules of the electricity spot price,a short-term electricity price forecasting method based on ARMAX model with skewness and kurtosis is proposed,in which the sine function and Gram-Charlier series expansion of the normal density function are used to describe the multicycle,skewness and kurtosis.So the skewness,kurtosis,multi-cycle and non-linear relationship among load and spot price can be fully taken into account.The numerical example based on the historical data of the PJM market shows that the model can reflect the features of electricity price better,and hold less computational cost,parsimonious scale of estimated parameters and high practical application value.