电力市场中的边际电价预测

在分析了系统边际价格(SMP)形成机理和影响因素的基础上，分别提出了基于累计式自回归滑动平均模型（ARIMA）和人工神经网络（ANN）的SMP预测方法，在这2种方法中都引入了市场供求指数(SDI)作为影响SMP的因素。通过对某省级发电市场真实数据的仿真结果表明，在引入SDI后，ARIMA模型和ANN模型的预测精度都得到了提高；同时，ANN模型比ARIMA模型更易于处理多种市场因素，若在模型中考虑更多的市场因素，则SMP预测的精度可进一步提高。

SYSTEM MARGINAL PRICE FORECASTING IN ELECTRICITY MARKET

The characteristics of system marginal price SMP are discussed according to economic theory.After that,two basic forecasting models of SMP are proposed based on auto- regression integrated moving- average ARIMA and artificial neural networks ANN respectively.A new factor named supply- demand index SDI ,reflecting the market balance of supply and demand,is taken into consideration in both models.Numerical result shows that the precision of these forecasting models is greatly improved after introducing SDI into the models.It is concluded that ANN is more flexible than ARIMA to deal with market factors in such cases.Moreover,more accurate forecasting results can be reached by taking more market factors into consideration.