基于分时段多模型的短期电价预测

Short-term price forecast models based on period-decoupled price sequence

基于分时段时间序列多模型的短期电价预测方法对美国PJM电力市场2006年全年电价数据进行分析,预测2007年1月1日到7日的一周内每小时的电价,将全年电价数据按照时段划分为24个子序列(PJM电力市场电价是以小时出清),分别对每个时段的子序列建立不同的模型进行分析,算例的研究结果显示,平均绝对百分误差在10%以内,能够用于电力市场短期电价预测。

The method which is based on period-decoupled price sequence was used to analyze the price data of PJM electricity market of 2006 to predict the hourly price from Jan 1^th to 7^th of 2007. With the method, the full yearly data were divided into 24 sub-sequences by hour, and relevant models for each period were constructed. The calculation example shows that the MAPE is around 10% and the model can be applied for short-term price forecast.