带置信区间的短期电价预测方法

Short-term electricity price forecasting method with confidence interval

电力市场中的电价呈现出均值回复、异方差、多周期、跳跃和尖峰等特性,预测精度往往不理想。同时,单点预测值难以有效指导市场参与者选择交易策略,也无法为电力监管机构提供监管和调控的依据。提出了一种带置信区间的小波—季节短期电价预测方法,以小波多分辨率分析和季节模型为工具,得到带置信区间的电价预测结果。算例分别采用加州市场平稳时期和危机时期历史数据,预测下一日电价,说明模型是有效的。

Since the electricity prices in the power market are usually characterized by mean reversion, heteroscedasticity, muhicycle, jump, and peak, the forecasting accuracy is not satisfactory. In the meanwhile, the single point prediction can not effectively guide the market participants to select the appropriate transaction strategy or provide the power regulatory organizations with the basis for supervising and regulating. A short-term electricity price forecasting method with a confidence interval is proposed, which uses the wavelet multi-resolution analysis and season models to obtain the forecast result with a confidence interval. The historical data of both the stable and crisis periods of California market are used in the case study, and the model is proved to be effective in forecasting the next-day prices.