基于自回归条件异方差-反向传播网络模型的日前边际电价预测.pdfVIP

32 卷 8 期 电 网 技 术 Vol. 32 No. 8 2008 年4 月 Power System Technology Apr. 2008 文章编号：1000-3673 （2008 ）08-0063-04 中图分类号：F407.2 文献标识码：A 学科代码：790 4940 基于自回归条件异方差-反向传播网络模型的 日前边际电价预测 帆，沈 炯，刘西陲 东南大学 能源与环境学院，江苏省 南京市 210096 ） Day-Ahead Marginal Price Forecasting Based on Autoregressive Conditional Heteroskedasticity-Back Propagation Network Model YU Fan SHEN Jiong LIU Xi-chui （School of Energy and Environment Southeast University Nanjing 210096 Jiangsu Province China ） ABSTRACT: An improved neural network model based on 项重要工作就是通过申报电量电价曲线参与市场 autoregressive conditional heteroskedasticity (ARCH) analysis 竞争。准确地预测出系统边际电价有助于经济独立 is proposed for day-ahead electricity market. Firstly, by use of 的各发电公司确定合适的报价策略，提高自身的经 ARCH analysis the conditional variance of marginal price 济效益。目前我国电力市场处于试运行阶段，电厂 series is obtained; then taking the conditional variance as the 报价数据和电力系统边际电价数据较少，可以借鉴 risk index of price fluctuation, an ARCH-back propagation 其他国家的经验和数据来开展这方面的研究。为 networks (BPN) model, which is based on historical prices, historical loads and conditional variances of historical prices, is 此，本文采用来自于已经成熟运行多年的美国PJM built, and by use of the built model the day-ahead marginal 市场的数据。 prices of Pennsylvania-New Jersey-Maryland (PJM) electricity 电价受经济、气象、电力系统运行情况、市场 market in United States are forecasted. Forecasting results 运行规则以及诸多人为因素的影响，准确预测电价 show that by means of leading in ARCH the forecasting 具有很大的困难，因此，诸多学者对边际电价的预 accuracy of traditional BPN can be effectively improved. 测算法进行了大量研究。文献[1]提出用时间序列方 KEY WORDS: electricity market ；day-ahead marginal price ； 法对西班牙电力市场的日前电价进行预测，取得了 forecasting ；autore