Bootstrap方法在基于ARMA-GARCH模型的电力市场价格区间预测中的应用(英文)  

Bootstrap prediction intervals for ARMA-GARCH based electricity market price forecasting

作　　者：陈霞[1] 董朝阳[1] 王殿辉 

机构地区：[1]昆士兰大学信息技术与电机工程学院,澳大利亚布里斯班QLD4067 [2]拉筹伯大学计算机科学与工程系,澳大利亚墨尔本Vie 3086

出　　处：《电力科学与技术学报》2008年第3期3-11,共9页Journal of Electric Power Science And Technology

摘　　要：当今开放式电力市场是一个基于竟价拍卖形式的市场,市场清算价格(market clearing price-MCP)主要受变化的供需平衡、市场参与者的竞价策略以及电力企业已经签订的双边买卖合同等因素影响,现有的价格预测技术只集中在如何提高价格预测准确度上,而忽视了对预测本身不确定性的度量;其中ARMA-GARCH价格预测模型的前提是假设预测误差符合正态分布,而实际电力市场价格非线性波动使得正态分布的假设完全不成立.针对这一现象,将不依赖任何分布假设的Bootstrap技术应用于基于ARMA-GARCH模型的电力市场清算价格区间预测,并对澳大利亚电力市场的真实数据进行分析评测,精度高,可靠性好.The deregulated electricity market is an auction market in which the market clearing price （MCP） is heavily affected by the nonstationary supply/demand balance, biding strategies of market participants, bilateral contracts the companies have signed etc. Existing price forecasting techniques have been devoted to finding the＂ best＂ point estimates rather than quantifying the predictive uncertainty. Interval/density forecasts are not new for most classic time series models, such ARMA-GARCH, where prediction intervals can be estimated analytically under normal distribution assumption. However, non-linearity nature of electricity price volatility makes the results unreliable. In this paper, we propose a bootstrap method for constructing prediction intervals for ARMA-GARCH based MCP forecasting process which does not reply on any distribution assumption. The Australian National Electricity Market （NEM） is selected as the primary case market for the evaluation purpose. Results indicate the bootstrap method is with high reliability and precision.

关 键 词：BOOTSTRAP 区间预测 GARCH 电力市场清算价格 

分 类 号：TM715[电气工程—电力系统及自动化] F407.61[经济管理—产业经济]