GMDH和变结构协整理论在电力负荷预测中的应用  

Application of GMDH and variable co-integration theory in power load forecasting

作　　者：顾洁[1] 储琳琳[2] 张宇俊[2] 施伟国[2] 

机构地区：[1]上海交通大学电气工程系,上海200240 [2]上海市电力公司市南供电公司,上海200233

出　　处：《电力系统保护与控制》2010年第22期80-85,共6页Power System Protection and Control

摘　　要：鉴于未考虑时间序列的非平稳性所得到的拟合方程可能出现'伪回归'而失去准确预测的能力,引入了协整理论与误差修正模型。考虑地区产业结构发生变化将导致地区用电量时间序列发生结构突变,影响平稳性检验结果的可信度,又引入了GMDH理论,自动搜索并确定结构突变点,以改进经典结构突变理论需要事先获知一些重大波动信息等主观性影响。结合上述两点,构建了基于GMDH理论的参数变结构协整模型。算例证明了该模型在电力系统的负荷预测中的适用性以及在处理'伪回归'和'样本数据结构突变'问题方面的有效性。The co-integration theory and Error Correction Model （ECM） are introduced in this paper to overcome the problem of spurious regression within the fitted equation which is caused by the neglect of time series nonstationarity. Taking into account the regional industrial structure changes leading to structural mutations of regional electricity consumption time series, and affecting the credibility of the results of stationarity tests, the Group Method of Data Handling （GMDH） theory has been introduced here to search and determine the structure of mutation points automatically, thus to avoid the subjective effects of classic structural change theory such as forecasting certain important fluctuation information. Combining the two points mentioned above, an improved variable co-integration of parameters model is established based on the theory of GMDH. The results of the example have demonstrated the effectiveness of the model to be used in load forecast area and to deal with the problem of ＂spurious regression＂ and ＂the structural mutations of sample data＂.

关 键 词：参数变结构协整 结构突变 数据处理组合方法(GMDH) 中长期电力负荷预测 误差修正模型(ErrorCorrection Model) 

分 类 号：TM715[电气工程—电力系统及自动化]