中国CO_2排放量变权组合预测研究

Variable Weight Combination Forecast of CO_2 Emission in China

从区域CO2排放量预测问题的离散灰色、非线性等特征出发,建立了基于离散灰色预测模型和广义回归神经网络模型的变权组合预测模型。其中,广义回归神经网络模型具有很强的非线性映射能力和柔性网络结构以及高度的容错性和鲁棒性,适用于解决非线性问题,有利于提高预测的准确性;在组合变权系数确定上,采用了等维递补多项式拟合方法,提高组合预测的拟合精度。以我国1990～2010年CO2排放量的测算数据以及同期的人口数量、GDP和能源消耗总量数据为基础,对未来7年我国CO2排放量进行了预测。

The variable weight combination forecast model is constructed based on discrete gray model（DGM） and generalized regression neural network（GRNN）,as well as the characteristics of regional CO2 emissions prediction,such as dispersed gray and nonlinearity.The generalized regression neural network model has a strong nonlinear mapping ability,flexible network structure,and a high degree of fault tolerance and robustness,suitable for solving nonlinear problems and improving the prediction accuracy.As to the determination of the variable weight,the equivalent dimensions additional polynomial fitting is adopted in order to improve the degree of fitting.Finally,on the basis of the estimated CO2 emissions of China during 1990～2010,the coetaneous population,GDP and total energy consumption,the CO2 emissions in China have been predicted for the next seven years.