【摘要】 对用电量、GDP变量系统建立了向量自回归模型,使用Johansen-juselius方法分析2个序列之间的协整关系,确定了协整向量,利用误差纠正项将无约束向量自回归模型修正为向量误差纠正模型,用极大似然估计参数完成模型的参数估计。从理论层面分析了2个变量系统的长期均衡关系和短期波动的误差纠正机制。最后运用该模型对江苏省季度数据进行预测分析。基于协整理论的负荷预测方案有效地避免了传统建模中出现的伪回归隐忧,实际算例表明,模型的预测效果是满意的。更多还原

【Abstract】 The theory of cointegration was introduced to the domain of load forecasting, which is a new branch of modern econometrics. The VAR model of bivariate system of seasonally adjusted power consumption and GDP were proposed. The mechanism of co-integration was discussed by Johansen-juselius method. The co-integration vector was presented and VAR model was adjusted to VECM. Using MLE, VECM was specified. The long term equilibrium and the mechanism of error correction in short term in this model were analyzed. Finally, the power consumption and GDP of Jiangsu province were forecasted by VEC model. The spurious regression can be avoided with the help of the method based on co-integration, and the results show that forecasting performance is satisfied.更多还原